PRIORITY INFORMATION

[0001] The present international application claims priority to United States Application Number 16/713,714 having a filing date of December 13, 2019, which is a continuation-in-part of United States Application Number 16/219,476 having a filing date of December 13, 2018, which claims the benefit of United States Provisional Application Serial Number 62/814,585 having a filing date of March 6, 2019. Applicants claim priority to and benefit of all such applications and incorporate all such applications herein by reference.

FIELD OF THE INVENTION

[0002] The present invention relates to computer-based reasoning systems and more specifically to synthetic data in computer-based reasoning systems.

BACKGROUND

[0003] Computer-based reasoning systems can be used to predict outcomes based on input data. For example, given a set of input data, a regression-based machine learning system can predict an outcome or make a decision. Computer-based reasoning systems will likely have been trained on much training data in order to generate its reasoning model. It will then predict the outcome or make a decision based on the reasoning model. [0004] One of the hardest problems for computer-based reasoning systems is, however, the acquisition of training data. Some systems may require millions or more sets of training data in order to properly train a system. Further, even when the computer-based reasoning system has enough data to use to train the computer-based reasoning system, that data may not be anonymous or anonymized in a way that satisfies user expectation, terms of service, etc. Other systems require the right sampling of training data. For example, even though a pump may spend 99% of its time in proper operating modes with similar data, a computer-based reasoning system to control it may need significantly more training on the potential failure scenarios with unusual data that comprise the other 1% of the operation time. Additionally, the training data may not be appropriate for use in reinforcement learning because significant amounts of data may be required in certain parts of the knowledge space or because the high costs associated with acquiring data such that the sampling process must be very selective.

[0005] The techniques herein overcome these issues.

[0006] The approaches described in this section are approaches that could be pursued, but not necessarily approaches that have been previously conceived or pursued. Therefore, unless otherwise indicated, it should not be assumed that any of the approaches described in this section qualify as prior art merely by virtue of their inclusion in this section.

SUMMARY

[0007] The claims provide a summary of the invention.

BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In the drawings:

[0009] FIG. 1 A, FIG. IB, and FIG. 1C are flow diagrams depicting example processes for synthetic data generation in computer-based reasoning systems.

[0010] FIG. 2 is a block diagram depicting example systems for synthetic data generation in computer-based reasoning systems.

[0011] FIG. 3 is a block diagram of example hardware for synthetic data generation in computer-based reasoning systems.

[0012] FIG. 4 is a flow diagram depicting example processes for controlling systems. DETAILED DESCRIPTION

[0013] In the following description, for the purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the present invention. It will be apparent, however, that the present invention may be practiced without these specific details. In other instances, well-known structures and devices are shown in block diagram form in order to avoid unnecessarily obscuring the present invention.

GENERAL OVERVIEW

[0014] The techniques herein provide for synthetic data generation based in computer-based reasoning systems. In some embodiments, the computer-based reasoning is a case-based reasoning system. As discussed elsewhere herein, computer-based reasoning systems need extensive, and often specific training data. It can be prohibitively expensive and time consuming to create such training data. Further, in numerous situations, including in the context of reinforcement learning, the specific training data needed to properly train a computer-based reasoning model may be difficult or impossible to obtain. Training is often needed on rare, but important, scenarios, which are costly, difficult, or dangerous to create, and, as such, that training data may not be available. The techniques herein use existing training data and, optionally, target surprisal to create synthetic training data. In some embodiments, conditions may also be applied to the creation of the synthetic data in order to ensure that training data meeting specific conditions is created. For undetermined features, a distribution for the feature among the training cases is determined, and a value for the feature is determined based on that distribution. In some embodiments, the values for some features are determined based at least in part on the conditions or condition requirements that are placed on the synthetic data. In some embodiments, the features that have conditions on them are called“conditioned features.” As used herein, the term “undetermined features” encompasses its plain and ordinary meaning, including, but not limited to those features for which a value has not yet been determined, and for which there is no condition or conditional requirement. For example, in those embodiments or instances where there are no conditions on the synthesized data, all of the features may initially be undetermined features. After a value for a feature is determined, it is no longer an undetermined feature, but instead (as described herein) may be used to as part of determining subsequent undetermined features. [0015] In some embodiments, feature bounds associated with the feature can be used to generate the value for the feature (e.g., using the techniques discussed herein). As particular examples, the feature value may be sampled based on a uniform distribution, truncated normal, or any other bounded distribution, between the feature bounds. In some embodiments, if the feature bounds are not used to generate the values, then the techniques herein include checking feature bounds (if any have been specified, determined, or are otherwise known) of the value determined for the feature. In some embodiments, if feature bounds are known for a specific feature, and the value determined for that feature is outside the feature bounds, then corrective action can be taken. Corrective action can include one or more of re-determining a value for the feature, and optionally again checking the new value against the feature bounds, choosing a value within the range of the feature bounds, replacing the determined value with a new value determined based on a distribution just within the feature bounds (e.g., a uniform, normal, or other distribution within the feature bounds), and / or the like. As a particular example, feature bounds may be used for time series velocities (discussed elsewhere herein).

[0016] In many embodiments or contexts, it may be important for synthetic training data to differ from the existing training data. For example, it may be useful to have the synthetic data not contain identical data cases as the original training data or even data cases that are overly “similar” or“close” to original training data cases. As such, in the event that synthetic training data is identical or too similar to existing training data, the synthetic training case may be modified (e.g., resampled) and retested, or discarded. In some embodiments, the generated synthetic data may be compared against the at least a portion of the existing training data, and a determination may be made whether to keep the synthetic training case based on the distance of the synthetic training case to one or more elements of in the existing training data. For example, in some embodiments, each synthetic training case generated using the techniques herein are compared to the existing training data in order to determine whether it is overly“similar” to existing training data. Determining whether the synthetic data case is overly similarity to a training case may be accomplished, in some embodiments, by determining the shortest distance between the synthetic data case and the data cases in the training data. If the shortest distance is below a certain threshold, then the synthetic training case is deemed to be too“close” to an existing case. In the event that the synthetic training case is overly similar to an existing training case, the synthetic case may be discarded, or values for one or more features may be redetermined and then the modified synthetic data case may be tested for similarity to the synthetic training cases. Any appropriate measure, metric, or premetric discussed herein may be used to determine the“distance” or“closeness” of the synthetic case with other cases, including Euclidean distance, Minkowski distance, Damerau-Levenshtein distance, Kullback-Leibler divergence, 1 - Kronecker delta, cosine similarity, Jaccard index, Tanimoto similarity, and / or any other distance measure, metric, pseudometric, premetric, index, etc. Further, the distance measure may be based on all of the features of the cases, or on a subset of the features. For example, the distance measure may be based on a subset of features that, for example, are known to be combinable as identifiers of individuals. Additionally, in some embodiments, the closeness of synthetic data cases is determined as the synthetic data cases are generated, after many or all of the synthetic data cases are generated, or a combination of the two.

[0017] In some embodiments, the techniques may include determining the k-anonymity for synthetic data cases (e.g., either as the synthetic data cases are generated, after many or all the synthetic data cases are generated, or a combination of the two) (e.g., as part of determining validity 152, fitness 160, and / or similarity 160 in FIG. 1A, IB, and / or 1C). In some embodiments, determining the k-anonymity of a synthetic data case may include determining whether there are k or more training data cases that are“close” to the synthetic data case (e.g., within a threshold distance - e.g., a“similarity threshold”), and, if there are at least k training data cases that are“close”, keeping the synthetic data case because one would not be able to associate the synthetic data case with less than k possible training data cases. In some embodiments, if there is at least one training data case, but fewer than k training data cases, that are within the similarity threshold of the synthetic data case, then the synthetic training data case may be discarded, or values for one or more features may be redetermined and then the modified synthetic data case may be tested for similarity and / or k-anonymity to the synthetic training cases. In some embodiments, even if there are k or more training data cases that are within the similarity threshold distance of the synthetic data case, if one or more of the training data cases are within a closer threshold distance to the synthetic case, then the synthetic training data case may be discarded, or values for one or more features or data elements in a time series may be redetermined and then the modified synthetic data case may be tested for similarity and / or k- anonymity to the synthetic training cases. This may be useful, for example, to avoid having one of the k or more training data cases be overly similar to the synthetic data case such that the synthetic data case and one or more of the training data cases are or are nearly identical. For example, if a synthetic data case is identical to one of the training cases, even if there are k other training cases that are similar, it may still be useful to exclude that case.

[0018] The number k used in k anonymity may be set by the users of the system or be automatically set based on desired anonymity, and may be related or unrelated to the“k” used for kNN searching. Further, the number k can be any appropriate number, such as 1, 2, 9, 101, 1432, etc. The number k can be the expected value of the number of data elements, the minimum number of elements that could be associated with a given point, or some other generalization of expectation including harmonic mean and geometric mean.

[0019] In some embodiments, the distribution for a feature may be perturbed based on target surprisal. In some embodiments, the distribution for a feature may be perturbed based on a multiplicity of surprisal or confidence values, including surprisal or confidence related to model residuals and the similarity or distance to existing points. In some embodiments, generated synthetic data may be tested for fitness. In some embodiments, generated synthetic data may be used as a sampling process to obtain observations about unknown parts of the computer-based reasoning model and to update the model based on the new information obtained. In some embodiments, generated synthetic data may be used as a sampling process that conditions the requests to increase the likelihood of the system it is driving attaining a goal. Further, the generated synthetic data may be provided in response to a request, used to train a computer- based reasoning model, and / or used to cause control of a system.

EXAMPLE PROCESSES FOR SYNTHETIC DATA GENERATION

[0020] FIG. 1 A is a flow diagram depicting example processes for synthetic data generation in computer-based reasoning systems. In some embodiments, process 100 proceeds by receiving 110 a request for synthetic training data. For example, a system or system operator may request additional or different training data in order to train a computer-based reasoning that will be used to control a system. In some cases, the system or operator may request anonymous data that is similar to a current training data set (or different from, but still anonymized). In other cases, the system or operator may require more data than is in the current training data set, and therefore may request additional data to augment the current training data set. In some cases, synthetic data may be requested to direct sampling via a reinforcement learning process. The synthesized data (perhaps combined with original training data or by itself) may be used as part of a computer-based reasoning system to cause control of a system. Many controllable systems can be controlled with the techniques herein, such as controllable machinery, autonomous vehicles, lab equipment, etc. In some embodiments, the request for synthetic data may include a target surprisal for the target data. In some embodiments, if low target surprisal is requested, then the synthetic data may be close to and not differ much from existing data. If high target surprisal is requested, then the generated synthetic data may differ much from the existing data.

[0021] The request can be received 110 in any appropriate manner, such as via HTTP, HTTPS, FTP, FTPS, a remote procedure call, an API, a function or procedure call, etc. The request can be formatted in any appropriate way, including in a structured format, such as HTML, XML, or a proprietary format or in a format acceptable by the API, remote procedure call, or function or procedure call. As one example, the request may be received 110 by a training and analysis system 210 in the manner discussed above.

[0022] In some embodiments, optionally, the received 110 request for synthetic data may also include one or more conditions for the synthetic data. These conditions may be restrictions on the generated synthetic data. For example, if the synthetic data being generated is for a checkers game, a condition on the data may be that includes only moves that are part of a winning strategy, that survive for at least S moves without losing, and / or win within W moves. Another set of conditions on the synthetic training data may be a particular board layout (e.g., the starting checkers game state, the current checkers game state), etc.

[0023] When the received 110 request includes one or more conditions for the synthetic data, the closest cases to the conditions may be determined 120 as focal cases. In some embodiments, the closest cases to the conditions may be determined as the K nearest neighbors (KNN) for the conditions (e.g., the K cases that are“closest” to meeting the conditions). For example, if there are two features that have conditions, A and B, and the conditions are A=3 and B=5, then the KNN for the conditions would be those cases that are closest to meeting the conditions of A=3 and B=5. In some instances, if there are more than K cases that fully meet the condition (e.g., there are more than K cases that have feature values of A=3 and B=5, which scenario will be more common if the conditions are on features which are nominal or categorical), then K cases may be selected from those cases meeting the condition. These K cases may be selected from among those that fully meet the conditions can be done randomly, or using any appropriate technique, such as by looking at the surprisal of those cases and choosing the K with the highest (or lowest) surprisal, or all of the K cases may be used. K may be 1, 2, 3, 5, 10, 100, a percentage of the model, specified dynamically or locally within the model, or any appropriate number. For distance measurements discussed herein (e.g., for use with K nearest neighbors), any appropriate measure, metric, or premetric may be used, including Euclidean distance, Minkowski distance, Damerau-Levenshtein distance, Kullback-Leibler divergence, 1 - Kronecker delta, cosine similarity, Jaccard index, Tanimoto similarity, and / or any other distance measure, metric, pseudometric, premetric, index, etc.

[0024] The conditions may be any appropriate single, multiple, and / or combination of conditions. For example, individual values may be given for features (e.g., A=5 and B=3); ranges may be given (e.g., A>=5 and B<4); multiple values may be given (e.g., E=“cat”,“dog”, or“horse”); one or more combination can be given (e.g., [(A>1 and B<99) or (A=7 and E= “horse”)]).

[0025] The values for the conditioned features may be set or determined based on the corresponding values for the features in the focal cases (e.g., determined as the KNN of the conditions, as described above). For example, for each conditioned feature, the mean, mode, an interpolated or extrapolated value, most-often occurring value of the corresponding feature from among the focal cases may be chosen as the value for the feature in the synthetic data case. In some embodiments, the distribution of the values of the conditioned features in the focal cases may be calculated and a value may be chosen based on that distribution, which may include the maximum likelihood value, selection via random sampling, inverse distance weighting, kernel functions, or other function or learned metric. In some embodiments, the values for conditioned features are set to (or based on) the condition values (vs. the values for the conditioned feature in the focal cases as described above). For example, if the conditions are A=5 and B=3, then feature A may be set to the value 5 and feature B may be set to the value 3 regardless of the values of that feature in the focal cases.

[0026] When there are no conditions received 110 with the request for synthetic data, a random case may be selected as a focal case or a set of random cases may be selected as the set of focal cases. When there are no conditions, then, in some embodiments, the techniques begin by selecting a random case, selecting the first feature, or a random feature, or the next feature prioritized by some metric, importance, conviction, or ranking, and select the value from the selected case as the value of the feature in the synthetic data value. Then, the techniques may proceed as described. For example, in some embodiments, the value for a first feature (e.g., A=12) is chosen from the chosen case and then the KNN are determined 120. The KNN may be the K cases that are closest to having that value (e.g., A=12) are chosen as the focal cases.

Additionally, other values computed from the data that are related to surprisal, confidence, or distance may be used in the selection of the focal cases (e.g., preferring the values to be chosen from areas where there is insufficient data, or when combined with other surprisal metrics, to prefer values where there is not a lack of data but where the model residuals or uncertainty are high).

[0027] After the focal cases for the synthetic data have been determined 120 (whether or not based on received 110 conditions), then a first undetermined feature is selected 130. When there are no conditions, selecting 130 the first undetermined feature comprises selecting 130 one of the features from the randomly selected case that was not previously determined. When there are conditions on the synthetic data, then the conditioned features are first set based on the conditions and the focal cases that are KNN of the condition (as described elsewhere herein). After the first feature(s) have been determined (whether or not there are conditions), then the next (undetermined) feature may be selected. Selecting 130 which undetermined feature to determine next can be done in any appropriate manner, such as selecting randomly among the remaining undetermined features, choosing the feature with the highest or lowest conviction, etc.

[0028] The distribution of values for the undetermined feature is then determined 140. For example, the distribution may be assumed to be log normal, Laplace, Gaussian, normal, or any other appropriate distribution, and be centered, e.g., on the computed undetermined feature value or on the median or mode or weighted sample or selection (e.g., weighted by probability, inverse distance, frequency, or other measure of likelihood) of the values for the undetermined feature in the set of focal cases (or in the training data). The distribution for the feature can also be determined by parameterizing it via surprisal using the distribution’s entropy. For example, if a distribution has an error, s, with the error modeled as a gaussian distribution, and we know that the entropy of a sample from gaussian distribution is 1 / 2 log(2 n e s ² ), we can adjust the error parameter to match a specified level of surprisal for that feature when taking a sample of the feature as the synthesized value. Alternatively, surprisal may also be determined by measuring other types of information, such as Kullback-Leibler Divergence (“KL divergence” or “DivKL(x)”) or cross entropy, and a desired surprisal can be attained by adjusting the corresponding parameters for the distribution. Methods describing distance from a point as a probability can be used to map the surprisal to distance, and may include any relevant distribution. When synthesizing data for multiple features, each feature can be set to the same surprisal, or alternatively each feature can“use up” the surprisal budget for the synthetic data parameterizing each feature’s distribution with its own amount of surprisal, treating total surprisal of the synthesized data as a budget or goal for all of the features together. Some features may be accorded more surprisal than others, and therefore may“use up” more of the surprisal budget. In cases where higher surprisal is desired, distributions will typically be wider. In situations where lower surprisal is desired, distributions will typically be narrower. The relative surprisal accorded each feature may be set or determined in any appropriate manner, including assigning the relative amount of surprisal randomly, having the relative amounts set by a human operator, and / or setting them based on a particular measure or metric, such as having the features with the lowest (or highest) surprisal in the training data being accorded more of the surprisal budget. Extensive additional discussion of these techniques are given elsewhere herein.

[0029] The value of for the undetermined feature for the synthetic case may then be determined 150 based on the determined 140 distribution. Determining the value based on the determined 140 distribution comprises selecting a value (or sampling) randomly based on a random number and the determined 140 distribution. In some embodiments, this is performed via inverse transform sampling. As one example, a random value representing the 3 ^rd percentile of the distribution of the random variable would translate to a value in the 3 ^rd percentile of the distribution, and a uniformly generated random number may be transformed into the distribution by the inverse cumulative density function. In some embodiments, the distribution does not have a closed form solution to translate a uniformly chosen random number to a random number from the parameterized distribution, and techniques to generate the required random number include rejection sampling, the Box-Muller transform, and the Ziggurat algorithm. As denoted by the dotted line from determining 150 and selecting 130, the process 100 may continue to determine values for features until there are no more undetermined features. In order to determine 150 values for each subsequent undetermined feature in the synthetic data case, the already- determined feature values are used to determine 120 the K nearest neighbors (a new set of focal cases) to that set of already-determined values (e.g., all of the feature values set to that point). For example, if values A=3, B=5, and C=9.7 have already been set for the synthetic data case, either via conditioning or using process 100 (and value D is next to be determined), then the K nearest neighbors to the values for A, B, and C will be the new set of focal cases. Then the distribution (e.g., DistD) for that subsequent undetermined feature (e.g., feature D) is determined 140 for the new set of focal cases. A value for the subsequent undetermined feature (e.g., D) is the determined based on a random sampling of the distribution (e.g., DistD) determined for that feature. When all of the feature values have been determined 150, then the synthetic data case is complete.

[0030] In some embodiments, determining 150 a value for a nominal feature may include skipping the distribution determining 140 step, and instead determining the value for the feature differently based on the desired surprisal or conviction. For example, if the desired conviction is infinity or an arbitrarily large number (low surprisal), then the value for the nominal feature may be chosen so that the data case as a whole represents a data case in the original training data cases, thus representing unsurprising results. If conviction is closer to one (e.g., within a threshold value of one), then the distribution used to determine the nominal value may be a blend of the global residual (e.g., the probability of each nominal value in the set of original training data cases) and the local residual (e.g., the probability of each nominal value in the set of focal cases). Blending the local residual and global residual may take any appropriate form, such as weighting each of the local and global residual, and combined the two weighted results. The weights may be any appropriate weight, such as equal weights, or a weighting determined based on conviction. Further, the blending of local and global residual may be more continuous than just two samples, involving multiple samples of the distribution based on the conviction. If desired conviction is or approaches zero (high surprisal), then the value for the nominal may be chosen randomly among the possible nominal values (e.g., using a 1/N probability for each of N possible nominal values). When conviction is outside the thresholds to use the“infinity” conviction, the“one” conviction, and the“zero” conviction, then, in various embodiments, different or similar calculation methods may be used. For example, in some embodiments, if desired conviction is less than one half, then the calculation techniques associated with a conviction of zero may be used. In some embodiments, a nominal value can be chosen based on a weighting of the two techniques, optionally with the distance to the key conviction value being the weighting. For example, with a desired conviction of 0.5, the weighting between use of 1/N probability for each of N possible nominal values may be weighted 50% and the blending of global and local residuals may be weighted 50% and the value for the feature may be determined based on the weighted results.

[0031] In some embodiments, optionally, the synthetic data case can be tested for fitness 160. Testing the synthetic data case for fitness 160 can include any appropriate technique, including confirming that the synthetic data case meets any received 110 conditions, or whether it meets other criteria, such as a fitness score or function. The fitness score or function may be any appropriate function. In some embodiments, the fitness function depends on the domain of the synthetic data case and can be a measure of performance of the synthetic data case as compared to other data cases. For example, the fitness function may be a measure of speed, processing efficiency, or some other measure of performance. Further, the fitness function might be modified at random, to introduce additional variation. Further, as discussed herein, determining the fitness of the synthetic data case may include determining the k-anonymity, validity, and / or similarity of the synthetic data case.

[0032] In some embodiments, after the completion of determination and testing of the synthetic data case, optionally, more synthetic data may be generated, as indicated by the dashed line from 170 to 120 (in the case of FIG. 1 A and FIG. IB) and / or 121 (in the case of FIG. 1C). The decision to generate more synthetic data may be based on the received 110 request for synthetic data. For example, the received 110 request may indicate that a certain number (or at least a certain number) of synthetic data cases are to be generated, that cases are generated based on a threshold for surprisal or conviction for new cases (e.g., generate cases as long as the surprisal of each new case is at least beyond a certain threshold), and / or the like. The process will then proceed as discussed herein. In some embodiments, not depicted in FIG. 1 A, FIG. IB, or FIG. 1C, the process will return to receive 110 more requests for synthetic data before proceeding to determine and test more synthetic data case(s).

[0033] Upon completion of determination and testing of the synthetic data case, optionally, it can be provided 170 as synthetic data. For example, the synthetic data case may be provided 170 in response to the received 110 request for data. In some embodiments, multiple synthetic data cases may be created in response to receiving 110 the original request, and may be provided 170 in response to that request. Providing the synthetic data case(s) in response to the request can take any appropriate form, including having them sent via HTTP, HTTPS, FTP, FTPS, via an API, a remote procedure call, a function or procedure call, etc., and / or in response to one of the foregoing.

[0034] In some embodiments, optionally, after one or more synthetic data cases have been created, control of a controllable system can be caused 199 based at least in part on the synthetic data case(s) created using process 100. For example, not depicted in FIG. 1 A, a computer-based reasoning model may be trained based on the synthetic data case(s) (and / or other sets of synthetic data cases, the training cases, and or a combination of such cases, or a combination of (sub)sets of such cases, etc.), and that model may be used to control a controllable system.

Numerous examples of causing 199 control of a controllable system are discussed herein and include, manufacturing control, vehicle control, image labelling control, smart device control, federated system control, etc.

ADDITIONAL EXAMPLE PROCESSES FOR SYNTHETIC DATA GENERATION: VALUE CHECKING

[0035] FIG. IB is a flow diagram depicting example processes for synthetic data generation in computer-based reasoning systems. Similar numbers and text are used to describe similar actions in FIG. 1 A, FIG. IB, and FIG. 1C. The process 100 described in FIG. IB may include different and / or additional steps. For example, turning to FIG. IB, as indicated by the dotted line, after the value for a feature is determined 150, the techniques optionally include checking whether the determined 150 value is valid 152. In some embodiments, the determined 150 value is checked for validity 152 using feature information, such as feature bounds and / or k- anonymity. For example, if feature bounds are known for a specific feature, and the value determined 150 for that feature is outside the feature bounds, then corrective action can be taken. Not pictured in FIG. 1, feature bounds may be determined based on the model, the data (e.g., observed and / or determined bounds of the feature), provided by an operator (e.g., set bounds of the feature), and / or the like. In some embodiments, other feature information may also be used to check whether the value is valid 152. For example, if the feature is related to one or more other features via correlation, inversely, or via other known relationship, then the value for the feature can be checked for validity based on the values of the one or more other related features. As a more specific example, if a feature indicates that a distributable agricultural quantity may be fertilizer or seeds, then the distribution rate bounds may differ considerably based on the type of distributable commodity. Nitrogen may have a distribution rate between one and six pounds per one thousand square feet. Ryegrass seeds, on the other hand, may have a distribution rate between five and ten pounds per one thousand square feet. Further, each of these distribution rates may be influenced by geographic region (e.g., as another feature). As such, checking the validity 152 of a distribution rate value may depend on both the type of agricultural item being distributed (a first additional feature) as well as geographic region (a second additional feature).

[0036] If the determined 150 value is determined to be invalid 152, then corrective action can be performed. Corrective action can include one or more of re-determining a value for the feature (e.g., as described with respect to determining 150), and optionally again checking the validity 152 of the newly determined 150 value. This process may continue until a valid value is determined, and / or up to a certain number of attempts before performing a different type of corrective action. In some embodiments, either after first re-determining 150 a new value for the feature, or instead of re-determining 150 a new value for the feature, a new value for the feature can be determined using the feature information, such as feature range. For example, the new value may be determined using a distribution across the feature range (e.g., using a uniform distribution, a normal distribution, etc.). In some embodiments, the new value for the feature may instead be determined by capping it at the feature boundaries.

ADDITIONAL EXAMPLE PROCESSES FOR SYNTHETIC DATA GENERATION:

IDENTITY AND / OR SIMILARITY CHECKING

[0037] As another example of the additions and or changes to the techniques or process 100 depicted in FIG. IB and FIG. 1C (vs. FIG. 1 A), the techniques may include, in addition to or instead of optionally determining fitness 160 of a synthetic data case, determining similarity 160 of the synthetic data case with all or a portion of the training data. Further, in some

embodiments, even if not depicted in the figures, determining the validity 152, fitness 160, and / or similarity 160 of generated data may be performed as part of the same step. As indicated in FIG. 1 A. For example, in FIG. 1 A, validity, fitness, and / or similarity are all performed as part of determining fitness 160 of the generated data. In some embodiments, determining fitness, validity, and similarity may be included in a single term of validity, fitness, or similarity.

[0038] As noted above, in some embodiments, it may be important for synthetic training data to differ from the existing training data. As such, the generated data may be checked for similarity with the training data. This similarity test may be based on a distance measure of all and / or a subset of the features. For example, if a subset of features is considered particularly sensitive or important, the distance measure may be used based on that subset of features. For example, instead of, or in addition to, checking the distance of all features between the generated data and the training data, the distance of a subset of the features may be checked. Furthering the example, if three features have been determined or are known to be particularly determinative, then the distance or similarity of those features for the generated data may be determined (as compared to the training data).

[0039] Similarity may be determined using Euclidean distance, Minkowski distance, Damerau-Levenshtein distance, Kullback-Leibler divergence, cosine similarity, Jaccard index, Tanimoto similarity, and / or any other distance measure, metric, pseudometric, premetric, index, and / or the like. In the event that synthetic training data is identical or too similar to existing training data (e.g., as measured by one of the measures discussed above), the synthetic training case may be modified (e.g., resampled) and retested, discarded, and / or replaced.

[0040] One or more newly generated synthetic data cases may be tested against at least a portion of the training data. For example, in some embodiments, it may be important that newly generated cases differ from the all data in the training data. As such, newly generated data cases may be assessed for similarity with all data cases in the training data. If over-similarity is found, then the synthetic case may be modified (and retested), discarded, or replaced (and the replacement synthetic case tested for similarity). In some embodiments, it may only be important that the synthetic data case differ from a portion of the existing training data. As such, the newly generated data case may be tested for similarity to only that portion of the cases. As a more specific example, if portions of the training data have personal information and other portions do not (e.g., the latter may have been synthetically generated, and / or may be part of the public domain), then the test for similarity may only be with respect to the portion of concern - the former portion containing personal information. In some embodiments, determining 160 similarity (or fitness) of the generated or synthetic data case may include determining the k- anonymity of the data case (as discussed elsewhere herein).

[0041] As noted above, determining the similarity of the synthetic training cases to the existing training data may be important in embodiments where the synthetic data needs to exclude identical and / or merely similar synthetic data as compared to the existing training data. For example, if there is personally identifiable information and / or data usable to re-identify an individual in the existing training data, then it may be useful if the synthetic data differs enough from the existing training data in order to ensure that the synthetic data cannot be re-identified to persons in the existing training data. As a more specific example, if the synthetic training case were generated as identical to a training case in the existing training data for person X, then that synthetic training case would be re-identifiable to person X. In contexts where it is desirable that the synthetic data does not represent any person from the existing training data, then it may be beneficial to determine whether there is any synthetic data that can be re-identified and modify or discard that synthetic data. Further, whether or not the data includes personally identifiable information, ensuring that the synthetic data cases differ from the existing training data may still be desirable. For example, it may be desirable that the synthetic training data not represent any particular existing training data in order to ensure that a model can be made without including that training data. For example, a data owner may want to share data (e.g., data related to machine vibration analysis) with a collaborator company, but may not want to share data specific to their operation. As such, they may desire that the synthetic training data not be overly similar to their existing training data, but still be useful for the collaborator company. The same can be true for a portion of the data. For example, the data owner may want to ensure that data related to a particular driver (for self-driving vehicles) not be included in the synthetic training data. As such, that driver’s data may be checked against the synthetic training data for over similarity, and any overly-similar data may be modified or discarded.

ADDITIONAL EXAMPLE PROCESSES FOR SYNTHETIC DATA GENERATION

[0042] FIG. 1C is a flow diagram depicting example processes for synthetic data generation in computer-based reasoning systems. Similar numbers and text are used to describe similar actions in FIG. 1 A, FIG. IB, and FIG. 1C. The process 100 described in FIG. 1C may include different and / or additional steps. For example, turning to FIG. 1C, the process depicted in FIG. 1C includes determining 121 one or more initial case(s). In some embodiments, there may be one initial case, and this initial case may be used as the basis on which the synthetic data is generated. The initial case may be chosen randomly from among the cases in the training data; the values of the features for the initial case may be chosen at random (e.g., set as random values that satisfy the type (e.g., integer, positive number, nominal, etc.) of the feature, such as choosing an integer if the feature requires an integer); may be chosen at random based on the distribution of the features in the training data (as discussed elsewhere herein); may be chosen to match one or more (or all) features of a case in the training data; a combination of these, etc. In some embodiments, two or more seed cases may be chosen (e.g., based on one or more conditions, at random, etc.), and the initial case can be determined based on the two or more seed cases (e.g., averaging the values from among the cases for each feature, using a voting mechanism for feature values, choosing randomly among the feature values of the seed cases, and / or the like).

[0043] After the initial case is determined 121, a feature in the case is selected 130 to be replaced, which may also be termed“dropping” the feature from the case. Selecting a feature is described elsewhere herein, and can include selecting a feature at random, selecting a feature with the lowest or highest conviction (of any type), and or any other technique, including those discussed herein. Once a feature is selected 130, then a value for the feature may be determined 150. Determining 150 the value for the feature may be accomplished based on the determined 140 distribution for the feature in the training data (as described elsewhere herein) as a whole (sometimes referred to as the global model). In some embodiments, the value for the features may be determined 150 by determining 140 the distribution for the feature for the local model (e.g., the k nearest neighbors). For example, in some embodiments, the closest k nearest neighbors of the case with the feature dropped out may be determined and the value for the feature may be determined based on the corresponding values for the feature in the k nearest neighbors. The local model distribution may be determined 140 and used to determine 150 the value for the feature. In some embodiments, not depicted in FIG. 1C, the value of the feature for the synthetic case can be determined from the corresponding values from the k nearest neighbors in any appropriate manner (e.g., not based on the local model distribution), including averaging the values from among the cases for each feature, using a voting mechanism among the values from among the cases for each feature, choosing randomly among the values, and / or the like.

[0044] In some embodiments, the value for more than one feature may be dropped out and determined 140 and / or 150 at the same time. For example, two, three, seven, etc. features may be dropped out, and valued for those two, three, seven, etc. features may be determined based on a global model, a local model, etc. as described herein.

[0045] In some embodiments, features are dropped out, and new values are determined 140 and / or 150 iteratively until a termination condition is met, as indicated by the dashed line between determining 150 and selecting 130. For example, in some embodiments, features will be dropped out and redetermined a fixed number of times (e.g., if there are F features, then this might happen a multiple of F times (e.g., 6*F), or a fixed number of times, such as 10, 100,

2000, 10 ^L 7, etc., regardless of F). Further, features may be dropped out and added back in in an order that ensures that each feature is replaced the same number of times, or a similar number of times (e.g., continue until all features have been replaced six times or replace all features until each feature has been replaced at least six times, but no more than seven times). In some embodiments, the number of times that a feature is replaced is not measured against the number of times that other features are dropped out and replaced, and the techniques proceed irrespective of that measure.

[0046] In some embodiments, features are dropped out and added back in until there is below a threshold difference between the case before and after the feature was dropped out and the value for it redetermined. For example, the distance between the case before the feature(s) are dropped out and replaced and the case after feature value(s) are replaced can be determined. In some embodiments, if the distance is below a particular threshold, then the process of iteratively dropping and replacing values will terminate. In some embodiments, the distance of the last R replacements may be measured. For example, if the distance between the pre-replacement case and the post-replacement case for the last R replacements is below a particular threshold, then the dropping and replacing of values will terminate. The distance measure used may be any appropriate distance measure, including those discussed herein and Euclidean distance,

Minkowski distance, Damerau-Levenshtein distance, Kullback-Leibler divergence, cosine similarity, Jaccard index, Tanimoto similarity, and / or any other distance measure, metric, pseudometric, premetric, index, or the like.

[0047] In some embodiments, the dropping and replacement of feature values may continue until the case is within a threshold distance of a case in the training data. For example, if a case is below a certain threshold distance to a case in the training data, the dropping and replacement of feature values may be terminated. This may be beneficial when it is the case that having a close distance between the synthetic case and a case in the training data means that the synthetic case is sufficiently similar to cases in the training data. Further, in some embodiments, the dropping and replacement of feature values may only terminate if the distance is below a certain closeness threshold, but above a second, identity threshold. This approach may be beneficial when being closer than an identity threshold means that the synthetic case risks being overly similar, and possibly identifiable as, a case in the training data.

[0048] In some embodiments, features are dropped out and added back in until there is below a threshold distance between the value of a feature dropped out and that added back in. For example, if the replacement value of a feature is within a certain predefined distance of the value it is replacing, then the process of dropping and replacing values may stop. In some

embodiments, the differences of the values for the last R replacements are checked to determine if, together, they are below a particular threshold. If they are, then the process of dropping and replacing values may be terminated and process 100 may continue.

[0049] In some embodiments, features are dropped out and added back in until a

combination of conditions are met. For example, in some embodiments, a check may be made to ensure that both (or either) of the conditions of the case changing by less than a first threshold distance is met and there having been M iterations already performed.

EXAMPLE TIME SERIES EMBODIMENTS

[0050] Data cases are discussed extensively elsewhere herein. Further to those discussions, in some embodiments, data cases may include features related to time series information for the features. For example, a data case at time“tO” may include N features. In some embodiments, in addition to those N features from time tO, one or more time series features may also be included, such as values from t(-l), t(-2), etc. The time series features may be previous value(s) for feature(s), difference(s) between the current value(s) and previous value(s), interpolated value(s) of previous value(s) for a fixed lag, differences between the current and previous value(s) divided by the timestep or time delta to the previous value(s) (akin to a“velocity”), and / or higher derivatives such as dividing by time twice or taking the change in velocity

(acceleration). Further, in some embodiments, time (e.g., as a value that indicates when the data case was collected or as a value indicating the delta to the previous time step) may be included include as a feature, and / or unique IDs to each time series of data is included. Some

embodiments include feature weighting (discussed elsewhere herein), that allows inclusion of more than one approach to time series inclusion, and determination of what works best. [0051] As noted above, the previous values may be included for less than the N total features. For example, it may be known that some features do not or change little change over time (e.g., names of an individual may be assumed to not change over a small time period).

[0052] Time series features may be used in the same manner as other features. For example, the time series information for a data case may be used in distance measure(s), may be used to test k-anonymity, similarity, validity, closeness, etc. Further, in some embodiments, time series information may be generated as part of determining a synthetic data case.

REINFORCEMENT LEARNING AND OTHER ADDITIONAL EMBODIMENTS

[0053] In some embodiments, the techniques may be used for reinforcement learning. For example, each time a synthetic training case is created, then the set of training cases can be updated and new synthetic data can be generated based on the updated set of training cases. In some embodiments, the techniques herein are used for reinforcement learning. For

reinforcement learning, the outcome or goal feature(s) (e.g., the score of a game, or having a winning checkers match) are treated as conditioned inputs or features. For example, in the checkers example, the synthetic data case is generated with conditions of the current game board setup and where the move was part of a winning strategy. The“winning strategy” feature may have been set in the training data set. For example, once a game has been won, an outcome feature is set to either“winning” or“losing” for all moves that had been made in the game. As such, each move in a winning game has the outcome feature set to“winning” and each move in a losing game has outcome set to“losing.” As such, then the data is conditioned to pick only moves that are part of a winning game, that feature (outcome=“winning”) is used in the KNN calculation discussed elsewhere herein.

[0054] The reinforcement learning scenarios can also include ranges (like a score above, below, or within a certain threshold), and other criteria. For example, as discussed elsewhere herein, the techniques herein can be useful in reinforcement learning situations where synthetic data is needed on expensive, dangerous, and / or hard to reproduce scenarios. For example, if pipelines only fail (e.g., leak, explode, become clogged) 0.001% of the time, but training data is needed to train a computer-based reasoning system to detect when those scenarios are going to happen, the techniques herein can be used to synthesize training data for those rare cases. This allows additional training data for pipeline failure to be gathered without incurring the difficulty, danger, and cost of actual pipeline failures. In such an example, the failure of the pipeline could be one of the conditions on the synthetic data. So, as data is being generated, the focal cases determined 120 will be those associated with pipeline failure, and the subsequently generated features will represent the distribution of values of those features within the conditioned data.

[0055] In some embodiments, the techniques may be used to create synthetic data that replicates users, devices, etc. For example, data that is based on, is similar to user data (or device data, etc.) can be created using the techniques herein. Consider user data that cannot be used (because it is not anonymous) and where one would prefer not to anonymize the data. That data can be used to create synthetic user data. If the data includes personally identifiable information as features (e.g., name, SSN, etc.), those features could be assigned random values, and the rest of the features can be synthesized based on user data (and possibly conditions) using the techniques discussed herein. Alternatively, in some embodiments, features containing personally identifiable information could also be generated based on existing user data, but with very high surprisal, creating a much wider distribution than seen in the user data.

OVERVIEW OF SURPRISAL, ENTROPY, AND DIVERGENCE

[0056] Below is a brief summary of some concepts discussed herein. It will be appreciated that there are numerous ways to compute the concepts below, and that other, similar

mathematical concepts can be used with the techniques discussed herein.

[0057] Entropy (“H(x)”) is a measure of the average expected value of information from an event and is often calculated as the sum over observations of the probability of each observation multiple by the negative log of the probability of the observation.

^H ( ^x ) = - å i p(xi) * log p(xi)

[0058] Entropy is generally considered a measure of disorder. Therefore, higher values of entropy represent less regularly ordered information, with random noise having high entropy, and lower values of entropy represent more ordered information, with a long sequence of zeros having low entropy. If log2 is used, then entropy may be seen as representing the theoretical lower bound on the number of bits needed to represent the information in a set of observations. Entropy can also be seen as how much a new observation distorts a combined probability density or mass function of the observed space. Consider, for example, a universe of observations where there is a certain probability that each of A, B, or C occurs, and a probability that something other than A, B, or C occurs.

[0059] Surprisal (“I(x)”) is a measure of how much information is provided by a new event

Xi.

I(xi) = - log p(xi)

[0060] Surprisal is generally a measure of surprise (or new information) generated by an event. The smaller the probability of Xi, the higher the surprisal.

[0061] Kullback-Leibler Divergence (“KL divergence” or“DIVKL(X)”) is a measure of difference in information between two sets of observation. It is often represented as

DIVKL(X) = å i p(xi) * (log p(xi) - log q(xi)), where p(xi) is the probability of xi after xi has occurred, and q(xi) is the probability of xi before xi has occurred.

FAMILIARITY CONVICTION EXAMPLES

[0062] Conviction and contribution measures may be used with the techniques herein. In some embodiments, conviction measures may be related in various ways to surprisal, including conviction being related to the ratio of observed surprisal to expected surprisal. Various of the conviction and contribution measures are discussed herein, including familiarity conviction discussed next.

[0063] In some embodiments, it may be useful to employ conviction as measure of how much information the point distorts the model. To do so, one may define a feature information measure, such as familiarity conviction, such that a point’s weighted distance contribution affects other points’ distance contribution and compared to the expected distance contribution of adding any new point.

[0064] Definition 1. Given a point x E X and the set K of its k nearest neighbors, a distance function d : R ^{z c} Z R, and a distance exponent a, the distance contribution of x may be the harmonic mean [0065] Definition 2. Given a set of points X c R ^z for every x e X and an integer 1 < k < |X| one may define the distance contribution probability distribution, C of X to be the set

for a function f : X R that returns the distance contribution.

[0066] Note that if f(0) = ¥, special consideration may be given to multiple identical points, such as splitting the distance contribution among those points.

[0067] Remark 1. C may be a valid probability distribution. In some embodiments, this fact is used to compute the amount of information in C.

[0068] Definition 3. The point probability of a point xi,i = 1,2, ...,n may be

where the index i is assigned the probability of the indexed point’s distance contribution. One may denote this random variable L.

[0069] Remark 2. When points are selected uniformly at random, one may assume L is uniform when the distance probabilities have no trend or correlation.

[0070] Definition 4. The conviction of a point xi e X may be

where KL is the Kullback-Leibler divergence. In some embodiments, when one assumes L is uniform, one may have that the expected probability ^~ ».

PREDICTION CONVICTION EXAMPLES

[0071] In some embodiments, it is useful to employ conviction as a proxy for accuracy of a prediction. To do so, one may define another type of conviction such that a point’s weighted distance to other points is of primary importance and can be expressed as the information required to describe the position of the point in question relative to existing points.

[0072] Definition 5. Let x be the number of features in a model and n the number of observations. One may define the residual function of the training data X:

r: X R

r(x) = Ji(k,p),J2(k,p),...,Jf(k,p) (7)

[0073] Where Ji may be the residual of the model on feature i parameterized by the hyperparameters k and p evaluated on points near x. In some embodiments, one may refer to the residual function evaluated on all of the model data as GM. in some embodiments, the feature residuals may be calculated as mean absolute error or standard deviation.

[0074] In some embodiments, one can quantify the information needed to express a distance contribution <p(x) by moving to a probability. In some embodiments, the exponential distribution may be selected to describe the distribution of residuals, as it may be the maximum entropy distribution constrained by the first moment. In some embodiments, a different distribution may be used for the residuals, such as the Laplace, lognormal distribution, Gaussian distribution, normal distribution, etc.

[0075] The exponential distribution may be represented or expressed as:

[0076] We can directly compare the distance contribution and p-normed magnitude of the residual. This is because the distance contribution is a locally weighted expected value of the distance from one point to its nearest neighbors, and the residual is an expected distance between a point and the nearest neighbors that are part of the model. Given the entropy maximizing assumption of the exponential distribution of the distances, we can then determine the probability that a distance contribution is greater than or equal to the magnitude of the residual ||r(x)|| as:

[0077] We then convert the probability to self-information as:

which simplifies to:

[0078] As the distance contribution decreases, or as the residual vector magnitude increases, the less information may be needed to represent this point. One can then compare this to the expected value a regular conviction form, yielding a prediction conviction of:

El

(12)

⁷¹ n Kxy

where / is the self-information calculated for each point in the model.

ADDITIONAL PREDICTION CONVICTION EXAMPLES

[0079] In some embodiments, <p(x) may be the distance contribution of point x, and r(x) may be the magnitude of the expected feature residuals at point x using the same norm and same topological parameters as the distance contribution, putting both on the same scale.

[0080] The probability of both being less than the expected values may be:

[0081] The self-information of this, which may be the negative log of the probability

[0082] The prediction conviction then may be calculated as: which simplifies to

FEATURE PREDICTION CONTRIBUTION EXAMPLES

[0083] In some embodiments, another feature information measure, Feature Prediction Contribution, may be related Mean Decrease in Accuracy (MDA). In MDA scores are established for models with all the features M and models with each feature held out M-f„i = 1... x. The difference \M - M- _/ \ is the importance of each feature, where the result’s sign is altered depending on whether the goal is to maximize or minimize score.

[0084] In some embodiments, prediction information _c is correlated with accuracy and thus may be used as a surrogate. The expected self-information required to express a feature is given by:

and the expected self-information to express a feature without feature i is

[0086] One can now make two definitions:

[0087] Definition 6. The prediction contribution : of feature i is

[0088] Definition 7. The prediction conviction, pi^, of feature i is

SYNTHETIC DATA GENERATION EXAMPLES [0089] In some embodiments, prediction conviction may express how surprising an observation is. As such, one may, effectively, reverse the math and use conviction to generate a new sample of data for a given amount of surprisal. In some embodiments, generally, the techniques may randomly select or predict a feature of a case from the training data and then resample it.

[0090] Given that some embodiments include calculating conditioned local residuals for a part of the model, as discussed elsewhere herein, the techniques may use this value to parameterize the random number distribution to generate a new value for a given feature. In order to understand this resampling method, it may be useful to discuss the approach used by the Mann-Whitney test, a powerful and widely used nonparametric test to determine whether two sets of samples were drawn from the same distribution. In the Mann-Whitney test, samples are randomly checked against one another to see which is greater, and if both sets of samples were drawn from the same distribution then the expectation is that both sets of samples should have an equal chance of having a higher value when randomly chosen samples are compared against each other.

[0091] In some embodiments, the techniques herein include resampling a point by randomly choosing whether the new sample is greater or less than the other point and then draw a sample from the distribution using the feature’s residual as the expected value. In some embodiments, using the exponential distribution yields the double-sided exponential distribution (also known as the Laplace distribution), though lognormal and other

distributions may be used as well.

[0092] If a feature is not continuous but rather nominal, then the local residuals can populate a confusion matrix, and an appropriate sample can be drawn based on the probabilities for drawing a new sample given the previous value.

[0093] As an example, the techniques may be used to generate a random value of feature i from the model with, for example, no other conditions on it. Because the observations within the model are representative of the observations made so far, a random instance is chosen from the observations using the uniform distribution over all observations. Then the value for feature i of this observation is resampled via the methods discussed elsewhere herein.

[0094] As another example, the techniques may be used to generate feature j of a data element or case, given that, in that data element or case, features i e X have corresponding values x The model labels feature j conditioned by all x to find some value t. This new value t becomes the expected value for the resampling process described elsewhere herein, and the local residual (or confusion matrix) becomes the appropriate parameter or parameters for the expected deviation.

[0095] In some embodiments, the techniques include filling in the features for an instance by beginning with no feature values (or a subset of all the feature values) specified as conditions for the data to generate. The remaining features may be ordered randomly or may be ordered via a feature conviction value (or in any other manner described herein). When a new value is generated for the current feature, then the process restarts with the newly-set feature value as an additional condition on that feature.

PARAMETERIZING SYNTHETIC DATA VIA PREDICTION CONVICTION EXAMPLES

[0096] As discussed elsewhere, various embodiments use the double-sided exponential distribution as a maximum entropy distribution of distance in Lp space. One may then be able to derive a closed form solution for how to scale the exponential distributions based on a prediction conviction value. For example, a value, v, for the prediction conviction may be expressed as n M.

n= = ΐ(c) (13) which may be rearranged as

El

7(x)= (14)

v

Substituting in the self-information described elsewhere herein:

cp(x) El

(15)

\V(x)\\ v

p

In some embodiments, that the units on both sides of Equation 15 match. This may be the case in circumstances where he natural logarithm and exponential in the derivation of Equation 15 cancel out, but leave the resultant in nats. We can rearrange in terms of distance contribution as: If we let p= 0, which may be desirable for conviction and other aspects of the similarity measure, then we can rewrite the distance contribution in terms of its parameter, l with expected mean of This becomes

[0101] In some embodiments, due to the number of ways surprisal may be assigned or calculated across the features, various solutions may exist. However, unless otherwise specified or conditioned, embodiments may include distributing surprisal uniformly across the features, holding expected proportionality constant. In some embodiments, the distance contribution may become the mean absolute error for the exponential distribution, such as:

and solving for the l to parameterize the exponential distributions may result in:

v

l = (19)

ϊ rEl

In some embodiments, Equation 19, when combined with the value of the feature, may become the distribution by which to generate a new random number under the maximum entropy assumption of exponentially distributed distance from the value.

REINFORCEMENT LEARNING EXAMPLES

[0102] In some embodiments, the techniques can generate data with a controlled amount of surprisal, which may be a novel way to characterize the classic exploration versus exploitation trade off in searching for an optimal solution to a goal. Traditionally, pairing a means to search, such as Monte Carlo tree search, with a universal function approximator, such as neural networks, may solve difficult reinforcement learning problems without domain knowledge. Because the data synthesis techniques described herein utilize the universal function approximator model (kNN) itself, it enables the techniques to be use in a

reinforcement learning architecture that is similar and tightly coupled, as described herein.

[0103] In some embodiments, setting the conviction of the data synthesis to“1” (or any other appropriate value) yields a balance between exploration and exploitation. Because, in some embodiments, the synthetic data generation techniques described herein can also be conditioned, the techniques may condition the search on both the current state of the system, as it is currently observed, and a set of goal values for features. In some embodiments, as the system is being trained, it can be continuously updated with the new training data. Once states are evaluated for their ultimate outcome, a new set of features or feature values can be added to all of the observations indicating the final scores or measures of outcomes (as described elsewhere herein, e.g., in relation to outcome features). Keeping track of which observations belong to which training sessions (e.g., games) may be beneficial as a convenient way to track and update this data. In some embodiments, given that the final score or multiple goal metrics may already be in the kNN database, the synthetic data generation may allow querying for new data conditioned upon having a high score or winning conditions (or any other appropriate condition), with a specified amount of conviction.

[0104] In some embodiments, the techniques herein provide a reinforcement learning algorithm that can be queried for the relevant training data for every decision, as described elsewhere herein. The commonality among the similar cases, boundary cases, archetypes, etc. can be combined to find when certain decisions are likely to yield a positive outcome, negative outcome, or a larger amount of surprisal thus improving the quality of the model. In some embodiments, by seeking high surprisal moves, the system will improve the breadth of its observations.

TARGETED AND UNTARGETED TECHNIQUES FOR DETERMINING CONVICTION AND OTHER MEASURES

[0105] In some embodiments, any of the feature information measures, conviction or contribution measures (e.g., surprisal, prediction conviction, familiarity conviction, and / or feature prediction contribution and / or feature prediction conviction) may be determined using an“untargeted” and / or a“targeted” approach. In the untargeted approach, the measure (e.g., a conviction measure) is determined by holding out the item in question and then measuring information gain associated with putting the item back into the model.

Various examples of this are discussed herein. For example, to measure the untargeted conviction of a case (or feature), the conviction is measured in part based on taking the case (or feature) out of the model, and then measuring the information associated with adding the case (or feature) back into the model.

[0106] In order to determine a targeted measure, such as surprisal, conviction, or contribution of a data element (e.g., a case or a feature), in contrast to untargeted measures, everything is dropped from the model except the features or cases being analyzed (the “analyzed data element(s)”) and the target features or cases (“target data element(s)”). Then the measure is calculated by measure the conviction, information gain, contribution, etc. based on how well the analyzed data element(s) predict the target data element(s) in the absence of the rest of the model.

[0107] In each instance that a measure, such as a surprisal, conviction, contribution, etc. measure, is discussed herein, the measure may be determined using either a targeted approach or an untargeted approach. For example, when the term“conviction” is used, it may refer to targeted or untargeted prediction conviction, targeted or untargeted familiarity conviction, and / or targeted or untargeted feature prediction conviction. Similarly, when surprisal, information, and / or contribution measures are discussed without reference to either targeted or untargeted calculation techniques, then reference may be being made to either a targeted or untargeted calculation for the measure.

SYSTEMS FOR SYNTHETIC DATA GENERATION IN COMPUTER-BASED

REASONING SYSTEMS

[0108] FIG. 2 is a block diagram depicting example systems for synthetic data generation in computer-based reasoning systems. Numerous devices and systems are coupled to a network 290. Network 290 can include the internet, a wide area network, a local area network, a Wi-Fi network, any other network or communication device described herein, and the like. Further, numerous of the systems and devices connected to 290 may have encrypted communication there between, VPNs, and or any other appropriate communication or security measure. System 200 includes a training and analysis system 210 coupled to network 290. The training and analysis system 210 may be used for collecting data related to systems 250 - 258 and creating computer-based reasoning models based on the training of those systems. Further, training and analysis system 210 may perform aspects of process 100 and / or 400 described herein. Control system 220 is also coupled to network 290. A control system 220 may control various of the systems 250 - 258. For example, a vehicle control 221 may control any of the vehicles 250 - 253, or the like. In some embodiments, there may be one or more network attached storages 230, 240. These storages 230, 240 may store training data, computer-based reasoning models, updated computer-based reasoning models, and the like.

In some embodiments, training and analysis system 210 and / or control system 220 may store any needed data including computer-based reasoning models locally on the system.

[0109] FIG. 2 depicts numerous systems 250 - 258 that may be controlled by a control system 220 or 221. For example, automobile 250, helicopter 251, submarine 252, boat 253, factory equipment 254, construction equipment 255, security equipment 256, oil pump 257, or warehouse equipment 258 may be controlled by a control system 220 or 221.

EXAMPLE PROCESSES FOR CONTROLLING SYSTEMS

[0110] FIG. 4 depicts an example process 400 for controlling a system. In some embodiments and at a high level, the process 400 proceeds by receiving or receiving 410 a computer-based reasoning model for controlling the system. The computer-based reasoning model may be one created using process 100, as one example. In some embodiments, the process 400 proceeds by receiving 420 a current context for the system, determining 430 an action to take based on the current context and the computer-based reasoning model, and causing 440 performance of the determined action (e.g., labelling an image, causing a vehicle to perform the turn, lane change, waypoint navigation, etc.). If operation of the system continues 450, then the process returns to receive 420 the current context, and otherwise discontinues 460 control of the system. In some embodiments, causing 199 performance of a selected action may include causing 440 performance of a determined action (or vice-versa).

[0111] As discussed herein the various processes 100, 400, etc. may run in parallel, in conjunction, together, or one process may be a subprocess of another. Further, any of the processes may run on the systems or hardware discussed herein. The features and steps of processes 100, 400 could be used in combination and / or in different orders.

SELF-DRIVING VEHICLES

[0112] Returning to the top of the process 400, it begins by receiving 410 a computer- based reasoning model for controlling the system. The computer-based reasoning model may be received in any appropriate matter. It may be provided via a network 290, placed in a shared or accessible memory on either the training and analysis system 210 or control system 220, or in accessible storage, such as storage 230 or 240.

[0113] In some embodiments (not depicted in FIG. 4), an operational situation could be indicated for the system. The operational situation is related to context, but may be considered a higher level, and may not change (or change less frequently) during operation of the system. For example, in the context of control of a vehicle, the operational situation may be indicated by a passenger or operator of the vehicle, by a configuration file, a setting, and / or the like. For example, a passenger Alicia may select“drive like Alicia” in order to have the vehicle driver like her. As another example, a fleet of helicopters may have a

configuration file set to operate like Bob. In some embodiments, the operational situation may be detected. For example, the vehicle may detect that it is operating in a particular location (area, city, region, state, or country), time of day, weather condition, etc. and the vehicle may be indicated to drive in a manner appropriate for that operational situation.

[0114] The operational situation, whether detected, indicated by passenger, etc., may be changed during operation of the vehicle. For example, a passenger may first indicate that she would like the vehicle to drive cautiously (e.g., like Alicia), and then realize that she is running later and switch to a faster operation mode (e.g., like Carole). The operational situation may also change based on detection. For example, if a vehicle is operating under an operational situation for a particular portion of road, and detects that it has left that portion of road, it may automatically switch to an operational situation appropriate for its location (e.g., for that city), may revert to a default operation (e.g., a baseline program that operates the vehicle) or operational situation (e.g., the last used). In some embodiments, if the vehicle detects that it needs to change operational situations, it may prompt a passenger or operator to choose a new operational situation.

[0115] In some embodiments, the computer-based reasoning model is received before process 400 begins (not depicted in FIG. 4), and the process begins by receiving 420 the current context. For example, the computer-based reasoning model may already be loaded into a controller 220 and the process 400 begins by receiving 420 the current context for the system being controlled. In some embodiments, referring to FIG. 2, the current context for a system to be controlled (not depicted in FIG. 2) may be sent to control system 220 and control system 220 may receive 420 current context for the system.

[0116] Receiving 420 current context may include receiving the context data needed for a determination to be made using the computer-based reasoning model. For example, turning to the vehicular example, receiving 420 the current context may, in various embodiments, include receiving information from sensors on or near the vehicle, determining information based on location or other sensor information, accessing data about the vehicle or location, etc. For example, the vehicle may have numerous sensors related to the vehicle and its operation, such as one or more of each of the following: speed sensors, tire pressure monitors, fuel gauges, compasses, global positioning systems (GPS), RADARs, LiDARs, cameras, barometers, thermal sensors, accelerometers, strain gauges, noise/sound measurement systems, etc. Current context may also include information determined based on sensor data. For example, the time to impact with the closest object may be determined based on distance calculations from RADAR or LiDAR data, and / or may be determined based on depth-from- stereo information from cameras on the vehicle. Context may include characteristics of the sensors, such as the distance a RADAR or LiDAR is capable of detecting, resolution and focal length of the cameras, etc. Context may include information about the vehicle not from a sensor. For example, the weight of the vehicle, acceleration, deceleration, and turning or maneuverability information may be known for the vehicle and may be part of the context information. Additionally, context may include information about the location, including road condition, wind direction and strength, weather, visibility, traffic data, road layout, etc.

[0117] Referring back to the example of vehicle control rules for Bob flying a helicopter, the context data for a later flight of the helicopter using the vehicle control rules based on Bob’s operation of the helicopter may include fuel remaining, distance that fuel can allow the helicopter to travel, location including elevation, wind speed and direction, visibility, location and type of sensors as well as the sensor data, time to impact with the N closest objects, maneuverability and speed control information, etc. Returning to the stop sign example, whether using vehicle control rules based on Alicia or Carole, the context may include LiDAR, RADAR, camera and other sensor data, location information, weight of the vehicle, road condition and weather information, braking information for the vehicle, etc.

[0118] The control system then determined 430 an action to take based on the current context and the computer-based reasoning model. For example, turning to the vehicular example, an action to take is determined 430 based on the current context and the vehicle control rules for the current operational situation. In some embodiments that use machine learning, the vehicle control rules may be in the form of a neural network (as described elsewhere herein), and the context may be fed into the neural network to determine an action to take. In embodiments using case-based reasoning, the set of context-action pairs closest (or most similar) to the current context may be determined. In some embodiments, only the closest context-action pair is determined, and the action associated with that context-action pair is the determined 430 action. In some embodiments, multiple context-action pairs are determined 430. For example, the N“closest” context-action pairs may be determined 430, and either as part of the determining 430, or later as part of the causing 440 performance of the action, choices may be made on the action to take based on the N closest context-action pairs, where“distance” for between the current context can be measured using any appropriate technique, including use of Euclidean distance, Minkowski distance, Damerau- Levenshtein distance, Kullback-Leibler divergence, and / or any other distance measure, metric, pseudometric, premetric, index, or the like.

[0119] In some embodiments, the actions to be taken may be blended based on the action of each context-action pair, with invalid (e.g., impossible or dangerous) outcomes being discarded. A choice can also be made among the N context-action pairs chosen based on criteria such as choosing to use the same or different operator context-action pair from the last determined action. For example, in an embodiment where there are context-action pair sets from multiple operators in the vehicle control rules, the choice of which context-action pair may be based on whether a context-action pair from the same operator was just chosen (e.g., to maintain consistency). The choice among the top N context-action pairs may also be made by choosing at random, mixing portions of the actions together, choosing based on a voting mechanism, etc.

[0120] Some embodiments include detecting gaps in the training data and / or vehicle control rules and indicating those during operation of the vehicle (for example, via prompt and / or spoken or graphical user interface) or offline (for example, in a report, on a graphical display, etc.) to indicate what additional training is needed (not depicted in FIG. 4). In some embodiments, when the computer-based reasoning system does not find context“close enough” to the current context to make a confident decision on an action to take, it may indicate this and suggest that an operator might take manual control of the vehicle, and that operation of the vehicle may provide additional context and action data for the computer- based reasoning system. Additionally, in some embodiments, an operator may indicate to a vehicle that she would like to take manual control to either override the computer-based reasoning system or replace the training data. These two scenarios may differ by whether the data (for example, context-action pairs) for the operational scenario are ignored for this time period, or whether they are replaced.

[0121] In some embodiments, the operational situation may be chosen based on a confidence measure indicating confidence in candidate actions to take from two (or more) different sets of control rules (not depicted in FIG. 4). Consider a first operational situation associated with a first set of vehicle control rules (e.g., with significant training from Alicia driving on highways) and a second operational situation associated with a second set of vehicle control rules (e.g., with significant training from Carole driving on rural roads). Candidate actions and associated confidences may be determined for each of the sets of vehicle control rules based on the context. The determined 430 action to take may then be selected as the action associated with the higher confidence level. For example, when the vehicle is driving on the highway, the actions from the vehicle control rules associated with Alicia may have a higher confidence, and therefore be chosen. When the vehicle is on rural roads, the actions from the vehicle control rules associated with Carole may have higher confidence and therefore be chosen. Relatedly, in some embodiments, a set of vehicle control rules may be hierarchical, and actions to take may be propagated from lower levels in the hierarchy to high levels, and the choice among actions to take propagated from the lower levels may be made on confidence associated with each of those chosen actions. The confidence can be based on any appropriate confidence calculation including, in some embodiments, determining how much“extra information” in the vehicle control rules is associated with that action in that context.

[0122] In some embodiments, there may be a background or baseline operational program that is used when the computer-based reasoning system does not have sufficient data to make a decision on what action to take (not depicted in FIG. 4). For example, if in a set of vehicle control rules, there is no matching context or there is not a matching context that is close enough to the current context, then the background program may be used. If none of the training data from Alicia included what to do when crossing railroad tracks, and railroad tracks are encountered in later operation of the vehicle, then the system may fall back on the baseline operational program to handle the traversal of the railroad tracks. In some embodiments, the baseline model is a computer-based reasoning system, in which case context-action pairs from the baseline model may be removed when new training data is added. In some embodiments, the baseline model is an executive driving engine which takes over control of the vehicle operation when there are no matching contexts in the vehicle control rules (e.g., in the case of a context-based reasoning system, there might be no context- action pairs that are sufficiently“close”).

[0123] In some embodiments, determining 430 an action to take based on the context can include determining whether vehicle maintenance is needed. As described elsewhere herein, the context may include wear and / or timing related to components of the vehicle, and a message related to maintenance may be determined based on the wear or timing. The message may indicate that maintenance may be needed or recommended (e.g., because preventative maintenance is often performed in the timing or wear context, because issues have been reported or detected with components in the timing or wear context, etc.). The message may be sent to or displayed for a vehicle operator (such as a fleet management service) and / or a passenger. For example, in the context of an automobile with sixty thousand miles, the message sent to a fleet maintenance system may include an indication that a timing belt may need to be replaced in order to avoid a P percent chance that the belt will break in the next five thousand miles (where the predictive information may be based on previously-collected context and action data, as described elsewhere herein). When the automobile reaches ninety thousand miles and assuming the belt has not been changed, the message may include that the chance that the belt will break has increased to, e.g., P*4 in the next five thousand miles.

[0124] Performance of the determined 430 action is then caused 440. Turning to the vehicular example, causing 440 performance of the action may include direct control of the vehicle and / or sending a message to a system, device, or interface that can control the vehicle. The action sent to control the vehicle may also be translated before it is used to control the vehicle. For example, the action determined 430 may be to navigate to a particular waypoint. In such an embodiment, causing 440 performance of the action may include sending the waypoint to a navigation system, and the navigation system may then, in turn, control the vehicle on a finer-grained level. In other embodiments, the determined 430 action may be to switch lanes, and that instruction may be sent to a control system that would enable the car to change the lane as directed. In yet other embodiments, the action determined 430 may be lower-level (e.g., accelerate or decelerate, turn 4° to the left, etc.), and causing 440 performance of the action may include sending the action to be performed to a control of the vehicle, or controlling the vehicle directly. In some embodiments, causing 440 performance of the action includes sending one or more messages for interpretation and / or display. In some embodiments, the causing 440 the action includes indicating the action to be taken at one or more levels of a control hierarchy for a vehicle. Examples of control hierarchies are given elsewhere herein.

[0125] Some embodiments include detecting anomalous actions taken or caused 440 to be taken. These anomalous actions may be signaled by an operator or passenger, or may be detected after operation of the vehicle (e.g., by reviewing log files, external reports, etc.). For example, a passenger of a vehicle may indicate that an undesirable maneuver was made by the vehicle (e.g., turning left from the right lane of a 2-lane road) or log files may be reviewed if the vehicle was in an accident. Once the anomaly is detected, the portion of the vehicle control rules (e.g., context-action pair(s)) related to the anomalous action can be determined. If it is determined that the context-action pair(s) are responsible for the anomalous action, then those context-action pairs can be removed or replaced using the techniques herein.

[0126] Referring to the example of the helicopter fleet and the vehicle control rules associated with Bob, the vehicle control 220 may determine 430 what action to take for the helicopter based on the received 420 context. The vehicle control 220 may then cause the helicopter to perform the determined action, for example, by sending instructions related to the action to the appropriate controls in the helicopter. In the driving example, the vehicle control 220 may determine 430 what action to take based on the context of vehicle. The vehicle control may then cause 440 performance of the determined 430 action by the automobile by sending instructions to control elements on the vehicle.

[0127] If there are more 450 contexts for which to determine actions for the operation of the system, then the process 400 returns to receive 410 more current contexts. Otherwise, process 400 ceases 460 control of the system. Turning to the vehicular example, as long as there is a continuation of operation of the vehicle using the vehicle control rules, the process 400 returns to receive 420 the subsequent current context for the vehicle. If the operational situation changes (e.g., the automobile is no longer on the stretch of road associated with the operational situation, a passenger indicates a new operational situation, etc.), then the process returns to determine the new operational situation. If the vehicle is no longer operating under vehicle control rules (e.g., it arrived at its destination, a passenger took over manual control, etc.), then the process 400 will discontinue 460 autonomous control of the vehicle.

[0128] Many of the examples discussed herein for vehicles discuss self-driving automobiles. As depicted in FIG. 2, numerous types of vehicles can be controlled. For example, a helicopter 251 or drone, a submarine 252, or boat or freight ship 253, or any other type of vehicle such as plane or drone (not depicted in FIG. 2), construction equipment, (not depicted in FIG. 2), and / or the like. In each case, the computer-based reasoning model may differ, including using different features, using different techniques described herein, etc. Further, the context of each type of vehicle may differ. Flying vehicles may need context data such as weight, lift, drag, fuel remaining, distance remaining given fuel, windspeed, visibility, etc. Floating vehicles, such as boats, freight vessels, submarines, and the like may have context data such as buoyancy, drag, propulsion capabilities, speed of currents, a measure of the choppiness of the water, fuel remaining, distance capability remaining given fuel, and the like. Manufacturing and other equipment may have as context width of area traversing, turn radius of the vehicle, speed capabilities, towing / lifting capabilities, and the like.

IMAGE LABELLING

[0129] The techniques herein may also be used for image-labeling systems. For example, numerous experts may label images (e.g., identifying features of or elements within those images). For example, the human experts may identify cancerous masses on x-rays. Having these experts label all input images is incredibly time consuming to do on an ongoing basis, in addition to being expensive (paying the experts). The techniques herein may be used to train an image-labeling computer-based reasoning model based on previously-trained images. Once the image-labeling computer-based reasoning system has been built, then input images may be analyzed using the image-based reasoning system. In order to build the image labeling computer-based reasoning system, images may be labeled by experts and used as training data. Using the techniques herein, the surprisal and / or conviction of the training data can be used to build an image-labeling computer-based reasoning system that balances the size of the computer-based reasoning model with the information that each additional image (or set of images) with associated labels provides. Once the image-labelling computer- based reasoning is trained, it can be used to label images in the future. For example, a new image may come in, the image-labelling computer-based reasoning may determine one or more labels for the image, and then the one or more labels may then be applied to the image. Thus, these images can be labeled automatically, saving the time and expense related to having experts label the images.

[0130] In some embodiments, processes 100, 400 may include determining the surprisal and / or conviction of each image (or multiple images) and the associated labels or of the aspects of the computer-based reasoning model. The surprisal and / or conviction for the one or more images may be determined and a determination may be made whether to select or include the one or more images (or aspects) in the image-labeling computer-based reasoning model based on the determined surprisal and / or conviction. While there are more sets of one or more images with labels to assess, the process may return to determine whether more image or label sets should be included or whether aspects should be included and / or changed in the model. Once there are no more images or aspects to consider, the process can turn to controlling the image analysis system using the image-labeling computer-based reasoning.

[0131] In some embodiments, process 100 may determine (e.g., in response to a request) synthetic data for use in the image-labeling computer-based reasoning model. Based on a model that uses the synthetic data, the process can cause 199 control of an image-labeling system using process 400. For example, if the data elements are related to images and labels applied to those images, then the image-labeling computer-based reasoning model trained on that data will apply labels to incoming images. Process 400 proceeds by receiving 410 an image-labeling computer-based reasoning model. The process proceeds by receiving 420 an image for labeling. The image-labeling computer-based reasoning model is then used to determine 430 labels for the input image. The image is then labeled 440. If there are more 450 images to label, then the system returns to receive 410 those images and otherwise ceases 460. In such embodiments, the image-labeling computer-based reasoning model may be used to select labels based on which training image is“closest” (or most similar) to the incoming image. The label(s) associated with that image will then be selected to apply to the incoming image.

MANUFACTURING AND ASSEMBLY

[0132] The processes 100, 400 may also be used for manufacturing and / or assembly.

For example, conviction can be used to identify normal behavior versus anomalous behavior of such equipment. Using the techniques herein, a crane (e.g., crane 255 of FIG. 2), robot arm, or other actuator is attempting to“grab” something and its surprisal is too high, it can stop, sound an alarm, shutdown certain areas of the facility, and/or request for human assistance. Anomalous behavior that is detected via conviction among sensors and actuators can be used to detect when there is some sort breakdown, unusual wear or mechanical or other malfunction, etc. It can also be used to find damaged equipment for repairs or buffing or other improvements for any robots or other machines that are searching and correcting defects in products or themselves (e.g., fixing a broken wire or smoothing out cuts made to the ends of a manufactured artifact made via an extrusion process). Conviction can also be used for cranes and other grabbing devices to find which cargo or items are closest matches to what is needed. Conviction can be used to drastically reduce the amount of time to train a robot to perform a new task for a new product or custom order, because the robot will indicate the aspects of the process it does not understand and direct training towards those areas and away from things it has already learned. Combining this with stopping ongoing actions when an anomalous situation is detected would also allow a robot to begin performing work before it is fully done training, the same way that a human apprentice may help out someone experienced while the apprentice is learning the job. Conviction can also inform what features or inputs to the robot are useful and which are not.

[0133] As an additional example in the manufacturing or assembly context, vibration data can be used to diagnose (or predict) issues with equipment. In some embodiments, the training data for the computer-based reasoning system would be vibration data (e.g., the output of one or more piezo vibration sensors attached to one or more pieces of

manufacturing equipment) for a piece of equipment along with diagnosis of an issue or error that occurred with the equipment. The training data may similarly include vibration data for the manufacturing equipment that is not associated with an issue or error with the

equipment. In subsequent operation of the same or similar equipment, the vibration data can be collected, and the computer-based reasoning model can be used to assess that vibration data to either diagnose or predict potential issues or errors with the equipment. For example, the vibration data for current (or recent) operation of one or more pieces of equipment, the computer-based reasoning model may be used to predict, diagnose, or otherwise determine issues or errors with the equipment. As a more specific example, a current context of vibration data for one or more pieces of manufacturing equipment may result in a diagnosis or prediction of various conditions, including, but not limited to: looseness of a piece of equipment (e.g., a loose screw), an imbalance on a rotating element (e.g., grime collected on a rotating wheel), misalignment or shaft runout (e.g., machine shafts may be out of alignment or not parallel), wear (e.g., ball or roller bearings, drive belts or gears become worn, they might cause vibration). As a further example, misalignment can be caused during assembly or develop over time, due to thermal expansion, components shifting or improper reassembly after maintenance. When a roller or ball bearing becomes pitted, for instance, the rollers or ball bearing will cause a vibration each time there is contact at the damaged area. A gear tooth that is heavily chipped or worn, or a drive belt that is breaking down, can also produce vibration. Diagnosis or prediction of the issue or error can be made based on the current or recent vibration data, and a computer-based reasoning model training data from the previous vibration data and associated issues or errors. Diagnosing or predicting the issues of vibration can be especially important where the vibration can cause other issues. For example, wear on a bearing may cause a vibration that then loosens another piece of equipment, which then can cause other issues and damage to equipment, failure of equipment, and even failure of the assembly or manufacturing process.

[0134] In some embodiments, techniques herein may determine (e.g., in response to a request) the surprisal and / or conviction of one or more data elements (e.g., of the manufacturing equipment) or aspects (e.g., features of context-action pairs or aspects of the model) to potentially include in the manufacturing control computer-based reasoning model. The surprisal and / or conviction for the one or more manufacturing elements may be determined and a determination may be made whether to select or include the one or more manufacturing data elements or aspects in the manufacturing control computer-based reasoning model based on the determined surprisal and / or conviction. While there are more sets of one or more manufacturing data elements or aspects to assess (e.g., from additional equipment and / or from subsequent time periods), the process may return to determine whether more manufacturing data elements or aspects sets should be included in the computer-based reasoning model. Once there are no more manufacturing data elements or aspects to consider for inclusion, the process can turn to controlling the manufacturing system using the manufacturing control computer-based reasoning system.

[0135] In some embodiments, process 100 may determine (e.g., in response to a request) synthetic data for use in the manufacturing control computer-based reasoning model. Based on a model using the synthetic data, causing 199 control of a manufacturing system may be accomplished by process 400. For example, if the data elements are related to manufacturing data elements or aspects, then the manufacturing control computer-based reasoning model trained on that data will cause control manufacturing or assemble. Process 400 proceeds by receiving 410 a manufacturing control computer-based reasoning model. The process proceeds by receiving 420 a context. The manufacturing control computer-based reasoning model is then used to determine 430 an action to take. The action is then performed by the control system (e.g., caused by the manufacturing control computer-based reasoning system). If there are more 450 contexts to consider, then the system returns to receive 410 those contexts and otherwise ceases 460. In such embodiments, the manufacturing control computer-based reasoning model may be used to control a manufacturing system. The chosen actions are then performed by a control system.

SMART VOICE CONTROL

[0136] The processes 100, 400 may be used for smart voice control. For example, combining multiple inputs and forms of analysis, the techniques herein can recognize if there is something unusual about a voice control request. For example, if a request is to purchase a high-priced item or unlock a door, but the calendar and synchronized devices indicate that the family is out of town, it could send a request to the person’s phone before confirming the order or action; it could be that an intruder has recorded someone’s voice in the family or has used artificial intelligence software to create a message and has broken in. It can detect other anomalies for security or for devices activating at unusual times, possibly indicating some mechanical failure, electronics failure, or someone in the house using things abnormally (e.g., a child frequently leaving the refrigerator door open for long durations). Combined with other natural language processing techniques beyond sentiment analysis, such as vocal distress, a smart voice device can recognize that something is different and ask, improving the person’s experience and improving the seamlessness of the device into the person’s life, perhaps playing music, adjusting lighting, or HVAC, or other controls. The level of confidence provided by conviction can also be used to train a smart voice device more quickly as it can ask questions about aspects of its use that it has the least knowledge about. For example:“I noticed usually at night, but also some days, you turn the temperature down in what situations should I turn the temperature down? What other inputs (features) should I consider?”

[0137] Using the techniques herein, a smart voice device may also be able to learn things it otherwise may not be able to. For example, if the smart voice device is looking for common patterns in any of the aforementioned actions or purchases and the conviction drops below a certain threshold, it can ask the person if it should take on a particular action or additional autonomy without prompting, such as“It looks like you’re normally changing the thermostat to colder on days when you have your exercise class, but not on days when it is cancelled; should I do this from now on and prepare the temperature to your liking?”

[0138] In some embodiments, processes 100, 400 may include determining (e.g., in response to a request) the surprisal and / or conviction of one or more data elements (e.g., of the smart voice system) or aspects (e.g., features of the data or parameters of the model) to potentially include in the smart voice system control computer-based reasoning model. The surprisal for the one or more smart voice system data elements or aspects may be determined and a determination may be made whether to include the one or more smart voice system data elements or aspects in the smart voice system control computer-based reasoning model based on the determined surprisal and / or conviction. While there are more sets of one or more smart voice system data elements or aspects to assess, the process may return to determine whether more smart voice system data elements or aspects sets should be included. Once there are no more smart voice system data elements or aspects to consider, the process can turn to controlling the smart voice system using the smart voice system control computer- based reasoning model.

[0139] In some embodiments, process 100 may determine (e.g., in response to a request) synthetic data for use in the smart voice computer-based reasoning model. Based on a model that uses the synthetic data, the process can cause 199 control of a smart voice system using process 400. For example, if the data elements are related to smart voice system actions, then the smart voice system control computer-based reasoning model trained on that data will control smart voice systems. Process 400 proceeds by receiving 410 a smart voice computer- based reasoning model. The process proceeds by receiving 420 a context. The smart voice computer-based reasoning model is then used to determine 430 an action to take. The action is then performed by the control system (e.g., caused by the smart voice computer-based reasoning system). If there are more 450 contexts to consider, then the system returns to receive 410 those contexts and otherwise ceases 460. In such embodiments, the smart voice computer-based reasoning model may be used to control a smart voice system. The chosen actions are then performed by a control system.

CONTROL OF FEDERARTED DEVICES

[0140] The processes 100, 400 may also be used for federated device systems. For example, combining multiple inputs and forms of analysis, the techniques herein can recognize if there is something that should trigger action based on the state of the federated devices. For example, if the training data includes actions normally taken and / or statuses of federated devices, then an action to take could be an often-taken action in the certain (or related contexts). For example, in the context of a smart home with interconnected heating, cooling, appliances, lights, locks, etc., the training data could be what a particular user does at certain times of day and / or in particular sequences. For example, if, in a house, the lights in the kitchen are normally turned off after the stove has been off for over an hour and the dishwasher has been started, then when that context again occurs, but the kitchen light has not been turned off, the computer-based reasoning system may cause an action to be taken in the smart home federated systems, such as prompting (e.g., audio) whether the user of the system would like the kitchen lights to be turned off. As another example, training data may indicate that a user sets the house alarm and locks the door upon leaving the house (e.g., as detected via geofence). If the user leaves the geofenced location of the house and has not yet locked the door and / or set the alarm, the computer-based reasoning system may cause performance of an action such as inquiring whether it should lock the door and / or set an alarm. As yet another example, in the security context, the control may be for turning on / off cameras, or enact other security measures, such as sounding alarms, locking doors, or even releasing drones and the like. Training data may include previous logs and sensor data, door or window alarm data, time of day, security footage, etc. and when security measure were (or should have been) taken. For example, a context such as particular window alarm data for a particular basement window coupled with other data may be associated with an action of sounding an alarm, and when a context occurs related to that context, an alarm may be sounded.

[0141] In some embodiments, processes 100, 400 may include determining the surprisal and / or conviction of one or more data elements or aspects of the federated device control system for potential inclusion in the federated device control computer-based reasoning model. The surprisal for the one or more federated device control system data elements may be determined and a determination may be made whether to select or include the one or more federated device control system data elements in the federated device control computer-based reasoning model based on the determined surprisal and / or conviction. While there are more sets of one or more federated device control system data elements or aspects to assess, the process may return to determine whether more federated device control system data elements or aspect sets should be included. Once there are no more federated device control system data elements or aspects to consider, the process can turn to controlling the federated device control system using the federated device control computer-based reasoning model.

[0142] In some embodiments, process 100 may determine (e.g., in response to a request) synthetic data for use in the federated device computer-based reasoning model. Based on a model that uses the synthetic data, the process can cause 199 control of a federated device system using process 400. For example, if the data elements are related to federated device system actions, then the federated device control computer-based reasoning model trained on that data will control federated device control system. Process 400 proceeds by receiving 410 a federated device control computer-based reasoning model. The process proceeds by receiving 420 a context. The federated device control computer-based reasoning model is then used to determine 430 an action to take. The action is then performed by the control system (e.g., caused by the federated device control computer-based reasoning system). If there are more 450 contexts to consider, then the system returns to receive 410 those contexts and otherwise ceases 460. In such embodiments, the federated device control computer- based reasoning model may be used to control federated devices. The chosen actions are then performed by a control system.

CONTROL AND AUTOMATION OF EXPERIMENTS

[0143] The processes 100, 400 may also be used to control laboratory experiments. For example, many lab experiments today, especially in the biological and life sciences, but also in agriculture, pharmaceuticals, materials science and other fields, yield combinatorial increases, in terms of numbers, of possibilities and results. The fields of design of experiment, as well as many combinatorial search and exploration techniques are currently combined with statistical analysis. However, conviction-based techniques such as those herein can be used to guide a search for knowledge, especially if combined with utility or fitness functions. Automated lab experiments (including pharmaceuticals, biological and life sciences, material science, etc.) may have actuators and may put different chemicals, samples, or parts in different combinations and put them under different circumstances.

Using conviction to guide the machines enables them to home in on learning how the system under study responds to different scenarios, and, for example, searching areas of greatest uncertainty (e.g., the areas with low conviction as discussed herein). Conceptually speaking, when the conviction or surprisal is combined with a fitness, utility, or value function, especially in a multiplicative fashion, then the combination is a powerful information theoretic approach to the classic exploration vs exploitation trade-offs that are made in search processes from artificial intelligence to science to engineering. Additionally, such a system can automate experiments where it can predict the most effective approach, homing in on the best possible, predictable outcomes for a specific knowledge base. Further, like in the other embodiments discussed herein, it could indicate (e.g., raise alarms) to human operators when the results are anomalous, or even tell which features being measured are most useful (so that they can be appropriately measured) or when measurements are not sufficient to characterize the outcomes. This is discussed extensively elsewhere herein. If the system has multiple kinds of sensors that have“costs” (e.g., monetary, time, computation, etc.) or cannot be all activated simultaneously, the feature entropies or convictions could be used to activate or deactivate the sensors to reduce costs or improve the distinguishability of the experimental results.

[0144] In the context of agriculture, growers may experiment with various treatments (plant species or varietals, crop types, seed planting densities, seed spacings, fertilizer types and densities, etc.) in order to improve yield and / or reduce cost. In comparing the effects of different practices (treatments), experimenters or growers need to know if the effects observed in the crop or in the field are simply a product of the natural variation that occurs in every ecological system, or whether those changes are truly a result of the new treatments.

In order to ameliorate the confusion caused by overlapping crop, treatment, and field effects, different design types can be used (e.g., demonstration strip, replication control or measurement, randomized block, split plot, factorial design, etc.). Regardless, however, of the type of test design type used, determination of what treatment(s) to use is crucial to success. Using the techniques herein to guide treatment selection (and possible design type) enables experimenters and growers to home in on how the system under study responds to different treatments and treatment types, and, for example, searching areas of greatest uncertainty in the“treatment space” (e.g., what are the types of treatments about which little is known?). Conceptually, the combination of conviction or surprisal with a value, utility, or fitness function such as yield, cost, or a function of yield and cost, become a powerful information theoretic approach to the classic exploration vs exploitation trade-offs that are made in search processes from artificial intelligence to science to engineering. Growers can use this information to choose treatments balancing exploitation (e.g., doing things similar to what has produced high yields previously) and exploration (e.g., trying treatments unlike previous ones, with yet-unknown results). Additionally, the techniques can automate experiments on treatments (either in selection of treatments, designs, or robotic or automated planting using the techniques described herein) where it can predict the most effective approach, and automatically perform the planting or other distribution (e.g., of fertilizer, seed, etc.) required of to perform the treatment. Further, like in the other embodiments discussed herein, it could indicate (e.g., raise alarms) to human operators when the results are anomalous, or even tell which features being measured are most useful or when

measurements are not useful to characterize the outcomes (e.g., and may possibly be discarded or no longer measured). If the system has types of sensors (e.g., soil moisture, nitrogen levels, sun exposure) that have“costs” (e.g., monetary, time, computation, etc.) or cannot be all collected or activated simultaneously, the feature entropies or convictions could be used to activate or deactivate the sensors to reduce costs while protecting the usefulness of the experimental results.

[0145] In some embodiments, processes 100, 400 may include determining (e.g., in response to a request) the surprisal and / or conviction of one or more data elements or aspects of the experiment control system. The surprisal for the one or more experiment control system data elements or aspects may be determined and a determination may be made whether to select or include the one or more experiment control system data elements or aspects in an experiment control computer-based reasoning model based on the determined surprisal and / or conviction. While there are more sets of one or more experiment control system data elements or aspects to assess, the process may return to determine whether more experiment control system data elements or aspects sets should be included. Once there are no more experiment control system data elements or aspects to consider, the process can cause 199 control of the experiment control system using the experiment control computer- based reasoning model.

[0146] In some embodiments, process 100 may determine (e.g., in response to a request) synthetic data for use in the experiment control computer-based reasoning model. Based on a model that uses the synthetic data, the process can cause 199 control of an experiment control system using process 400. For example, if the data elements are related to experiment control system actions, then the experiment control computer-based reasoning model trained on that data will control experiment control system. Process 400 proceeds by receiving 410 an experiment control computer-based reasoning model. The process proceeds by receiving 420 a context. The experiment control computer-based reasoning model is then used to determine 430 an action to take. The action is then performed by the control system (e.g., caused by the experiment control computer-based reasoning system). If there are more 450 contexts to consider, then the system returns to receive 410 those contexts and otherwise ceases 460. In such embodiments, the experiment control computer-based reasoning model may be used to control experiment. The chosen actions are then performed by a control system.

CONTROL OF ENERGY TRANSFER SYSTEMS

[0147] The processes 100, 400 may also be used for control systems for energy transfer. For example, a building may have numerous energy sources, including solar, wind, grid- based electrical, batteries, on-site generation (e.g., by diesel or gas), etc. and may have many operations it can perform, including manufacturing, computation, temperature control, etc. The techniques herein may be used to control when certain types of energy are used and when certain energy consuming processes are engaged. For example, on sunny days, roof- mounted solar cells may provide enough low-cost power that grid-based electrical power is discontinued during a particular time period while costly manufacturing processes are engaged. On windy, rainy days, the overhead of running solar panels may overshadow the energy provided, but power purchased from a wind-generation farm may be cheap, and only essential energy consuming manufacturing processes and maintenance processes are performed.

[0148] In some embodiments, processes 100, 400 may include determining (e.g., in response to a request) the surprisal and / or conviction of one or more data elements or aspects of the energy transfer system. The surprisal for the one or more energy transfer system data elements or aspects may be determined and a determination may be made whether to select or include the one or more energy transfer system data elements or aspects in energy control computer-based reasoning model based on the determined surprisal and / or conviction. While there are more sets of one or more energy transfer system data elements or aspects to assess, the process may return to determine whether more energy transfer system data elements or aspects should be included. Once there are no more energy transfer system data elements or aspects to consider, the process can turn to controlling the energy transfer system using the energy control computer-based reasoning model.

[0149] In some embodiments, process 100 may determine (e.g., in response to a request) synthetic data for use in the energy transfer computer-based reasoning model. Based on a model that uses the synthetic data, the process can cause 199 control of an energy transfer system using process 400. For example, if the data elements are related to energy transfer system actions, then the energy control computer-based reasoning model trained on that data will control energy transfer system. Process 400 proceeds by receiving 410 an energy control computer-based reasoning model. The process proceeds by receiving 420 a context. The energy control computer-based reasoning model is then used to determine 430 an action to take. The action is then performed by the control system (e.g., caused by the energy control computer-based reasoning system). If there are more 450 contexts to consider, then the system returns to receive 410 those contexts and otherwise ceases 460. In such embodiments, the energy control computer-based reasoning model may be used to control energy. The chosen actions are then performed by a control system.

HEALTH CARE DECISION MAKING, PREDICTION, AND FRAUD PROTECTION

[0150] The processes 100, 400 may also be used for health care decision making, prediction (such as outcome prediction), and fraud detection. For example, some health insurers require pre-approval, pre-certification, pre-authorization, and / or reimbursement for certain types of healthcare procedures, such as healthcare services, administration of drugs, surgery, hospital visits, etc. When analyzing pre-approvals, a health care professional must contact the insurer to obtain their approval prior to administering care, or else the health insurance company may not cover the procedure. Not all services require pre-approval, but many may, and which require it can differ among insurers. Health insurance companies may make determinations including, but not necessarily limited to, whether a procedure is medically necessary, whether it is duplicative, whether it follows currently-accepted medical practice, whether there are anomalies in the care or its procedures, whether there are anomalies or errors with the health care provider or professional, etc.

[0151] In some embodiments, a health insurance company may have many“features” of data on which health care pre-approval or reimbursement decisions are determined by human operators. These features may include diagnosis information, type of health insurance, requesting health care professional and facility, frequency and / or last claim of the particular type, etc. The data on previous decisions can be used to train the computer-based reasoning system. The techniques herein may be used to guide the health care decision making process. For example, when the computer-based reasoning model determines, with high conviction or confidence, that a procedure should be pre-approved or reimbursed, it may pre-approve or reimburse the procedure without further review. In some embodiments, when the computer- based reasoning model has low conviction re whether or not to pre-approve a particular procedure, it may flag it for human review (including, e.g., sending it back to the submitting organization for further information). In some embodiments, some or all of the rejections of procedure pre-approval or reimbursement may be flagged for human review.

[0152] Further, in some embodiments, the techniques herein can be used to flag trends, anomalies, and / or errors. For example, as explained in detail elsewhere herein, the techniques can be used to determine, for example, when there are anomalies for a request for pre-approval, diagnoses, reimbursement requests, etc. with respect to the computer-based reasoning model trained on prior data. When the anomaly is detected, (e.g., outliers, such as a procedure or prescription has been requested outside the normal range of occurrences per time period, for an individual that is outside the normal range of patients, etc.; and / or what may be referred to as“inliers” - or“contextual outliers,” such as too frequently (or rarely) occurring diagnoses, procedures, prescriptions, etc.), the pre-approval, diagnosis,

reimbursement request, etc. can be flagged for further review. In some cases, these anomalies could be errors (e.g., and the health professional or facility may be contacted to rectify the error), acceptable anomalies (e.g., patients that need care outside of the normal bounds), or unacceptable anomalies. Additionally, in some embodiments, the techniques herein can be used to determine and flag trends (e.g., for an individual patient, set of patients, health department or facility, region, etc.). The techniques herein may be useful not only because they can automate and / or flag pre-approval decision, reimbursement requests, diagnosis, etc., but also because the trained computer-based reasoning model may contain information (e.g., prior decision) from multiple (e.g., 10s, 100s, 1000s, or more) prior decision makers. Consideration of this large amount of information may be untenable for other approaches, such as human review.

[0153] The techniques herein may also be used to predict adverse outcomes in numerous health care contexts. The computer-based reasoning model may be trained with data from previous adverse events, and perhaps from patients that did not have adverse events. The trained computer-based reasoning system can then be used to predict when a current or prospective patient or treatment is likely to cause an adverse event. For example, if a patient arrives at a hospital, the patient’s information and condition may be assessed by the computer-based reasoning model using the techniques herein in order to predict whether an adverse event is probable (and the conviction of that determination). As a more specific example, if a septuagenarian with a history of low blood pressure is admitted for monitoring a heart murmur, the techniques herein may flag that patient for further review. In some embodiments, the determination of a potential adverse outcome may be an indication of one or more possible adverse events, such as a complication, having an additional injury, sepsis, increased morbidity, and / or getting additionally sick, etc. Returning to the example of the septuagenarian with a history of low blood pressure, the techniques herein may indicate that, based on previous data, the possibility of a fall in the hospital is unduly high (possibly with high conviction). Such information can allow the hospital to try to ameliorate the situation and attempt to prevent the adverse event before it happens.

[0154] In some embodiments, the techniques herein include assisting in diagnosis and / or diagnosing patients based on previous diagnosis data and current patient data. For example, a computer-based reasoning model may be trained with previous patient data and related diagnoses using the techniques herein. The diagnosis computer-based reasoning model may then be used in order to suggest one or more possible diagnoses for the current patient. As a more specific example, a septuagenarian may present with specific attributes, medical history, family history, etc. This information may be used as the input context to the diagnosis computer-based reasoning system, and the diagnosis computer-based reasoning system may determine one or more possible diagnoses for the septuagenarian. In some embodiments, those possible diagnoses may then be assessed by medical professionals. The techniques herein may be used to diagnose any condition, including, but not limited to breast cancer, lung cancer, colon cancer, prostate cancer, bone metastases, coronary artery disease, congenital heart defect, brain pathologies, Alzheimer's disease, and / or diabetic retinopathy.

[0155] In some embodiments, the techniques herein may be used to generate synthetic data that mimics, but does not include, previous patient data. This synthetic data generation is available for any of the uses of the techniques described herein (manufacturing, image labelling, self-driving vehicles, etc.), and can be particularly important in circumstances where using user data (such as patient health data) in a model may be contrary to policy or regulation. As discussed elsewhere herein, the synthetic data can be generated to directly mimic the characteristics of the patient population, or more surprising data can be generated (e.g., higher surprisal) in order to generate more data in the edge cases, all without a necessity of including actual patient data.

[0156] In some embodiments, processes 100, 400 may include determining (e.g., in response to a request) the surprisal and / or conviction of one or more data elements or aspects of the health care system. The surprisal or conviction for the one or more health care system data elements or aspects may be determined and a determination may be made whether to select or include the one or more health care system data elements or aspects in a health care system computer-based reasoning model based on the determined surprisal and / or conviction. While there are more sets of one or more health care system data elements or aspects to assess, the process may return to determine whether more health care system data elements or aspects should be included. Once there are no more health care system data elements or aspects to consider included in the model, the process can turn to controlling the health care computer-based reasoning system using the health care system computer-based reasoning model.

[0157] In some embodiments, process 100 may determine (e.g., in response to a request) synthetic data for use in the health care system computer-based reasoning model. Based on a model that uses the synthetic data, the process can cause 199 control of a health care computer-based reasoning system using process 400 For example, if the data elements are related to health care system actions, then the health care system computer-based reasoning model trained on that data will control the health care system. Process 400 proceeds by receiving 410 a health care system computer-based reasoning model. The process proceeds by receiving 420 a context. The health care system computer-based reasoning model is then used to determine 430 an action to take. The action is then performed by the control system (e.g., caused by the health care system computer-based reasoning system). If there are more 450 contexts to consider, then the system returns to receive 410 those contexts and otherwise ceases 460 In such embodiments, the health care system computer-based reasoning model may be used to assess health care decisions, predict outcomes, etc. In some embodiments, the chosen action(s) are then performed by a control system.

REAL ESTATE FUTURE VALUE AND VALUATION PREDICTION

[0158] The techniques herein may also be used for real estate value estimation. For example, the past values and revenue from real estate ventures may be used as training data. This data may include, in addition to value (e.g., sale or resale value), compound annual growth rate (“CAGR”), zoning, property type (e.g., multifamily, Office, Retail, Industrial), adjacent business and types, asking rent (e.g., rent per square foot (“sqft”) for each of Office, Retail, Industrial, etc. and / or per unit (for multifamily buildings), further, this may be based on all properties within the selected property type in a particular geography, for example), capitalization rate (or“cap rate” based on all properties within selected property type in a geography), demand (which may be quantified as occupied stock), market capitalization (e.g., an average modeled price per sqft multiplied by inventory sqft of the given property type and / or in a given geography), net absorption (net change in demand for a rolling 12 month period), net completions (e.g., net change in inventory sqft (Office, Retail, Industrial) or units (Multifamily) for a period of time, such as analyzed data element(s) rolling 12 month period), occupancy (e.g., Occupied sqft / total inventory sqft, 100% - vacancy%, etc.), stock (e.g., inventory square footage (Office, Retail, Industrial) or units (Multifamily), revenue (e.g., revenue generated by renting out or otherwise using a piece of real estate), savings (e.g., tax savings, depreciation), costs (e.g., taxes, insurance, upkeep, payments to property managers, costs for findings tenants, property managers, etc.), geography and geographic location (e.g., views of water, distance to shopping, walking score, proximity to public transportation, distance to highways, proximity to job centers, proximity to local universities, etc.), building characteristics (e.g., date built, date renovated, etc.), property characteristics (e.g., address, city, state, zip, property type, unit type(s), number of units, numbers of bedrooms and bathrooms, square footage(s), lot size(s), assessed value(s), lot value(s), improvements value(s), etc. - possibly including current and past values), real estate markets characteristics (e.g., local year-over-year growth, historical year-over-year growth), broader economic information (e.g., gross domestic product growth, consumer sentiment, economic forecast data), local economic information (e.g., local economic growth, average local salaries and growth, etc.), local demographics (e.g., numbers of families, couples, single people, number of working-age people, numbers or percentage of people with at different education, salary, or savings levels, etc.). The techniques herein may be used to train a real estate computer- based reasoning model based on previous properties. Once the real estate computer-based reasoning system has been trained, then input properties may be analyzed using the real estate reasoning system. Using the techniques herein, the surprisal and / or conviction of the training data can be used to build an real estate computer-based reasoning system that balances the size of the computer-based reasoning model with the information that each additional property record (or set of records) provides to the model.

[0159] The techniques herein may be used to predict performance of real estate in the future. For example, based on the variables associated discussed here, that are related, e.g., with various geographies, property types, and markets, the techniques herein may be used to find property types and geographies with the highest expected value or return (e.g., as CAGR). As a more specific example, a model of historical CAGR with asking rent, capitalization rate, demand, net absorption, net completions, occupancy, stock, etc. can be trained. That model may be used, along with more current data, to predict the CAGR of various property types and / or geographies over the coming X years (e.g., 2, 3, 5, or 10 years). Such information may be useful for predicting future value for properties and / or automated decision making. [0160] As another example, using the techniques herein, a batch of available properties may be given as input to the real estate computer-based reasoning systems, and the real estate computer-based reasoning system may be used to determine what properties are likely to be good investments. In some embodiments, the predictions of the computer-based reasoning system may be used to purchase properties. Further, as discussed extensively herein, explanations may be provided for the decisions. Those explanation may be used by a controllable system to make investment decisions and / or by a human operator to review the investment predictions.

[0161] In some embodiments, processes 100, 400 may include determining the surprisal and / or conviction of each input real estate data case (or multiple real estate data cases) with respect to the associated labels or of the aspects of the computer-based reasoning model. The surprisal and / or conviction for the one or more real estate data cases may be determined and a determination may be made whether to select or include the one or more real estate data cases in the real estate computer-based reasoning model based on the determined surprisal and / or conviction. While there are more sets of one or more real estate data cases to assess, the process may return to determine whether more real estate data case sets should be included or whether aspects should be included and / or changed in the model. Once there are no more training cases to consider, the process can turn to controlling predicting real estate investments information for possible use in purchasing real estate using the real estate computer-based reasoning.

[0162] In some embodiments, process 100 may determine (e.g., in response to a request) synthetic data for use in the real estate system computer-based reasoning model. Based on a model that uses the synthetic data, the process can cause 199 control of a real estate system, using, for example, process 400. For example, the training data elements are related to real estate, and the real estate computer-based reasoning model trained on that data will determined investment value(s) for real estate data cases (properties) under consideration. These investment values may be any appropriate value, such as CAGR, monthly income, resale value, income or resale value based on refurbishment or new development, net present value of one or more of the preceding, etc. In some embodiments, process 400 begins by receiving 410 a real estate computer-based reasoning model. The process proceeds by receiving 420 properties under consideration for labeling and / or predicting value(s) for the investment opportunity. The real estate computer-based reasoning model is then used to determine 430 values for the real estate under consideration. The prediction(s) for the real estate is (are) then made 440. If there are more 450 properties to consider, then the system returns to receive 410 data on those properties and otherwise ceases 460. In some

embodiments, the real estate computer-based reasoning model may be used to determine which training properties are“closest” (or most similar) to the incoming property or property types and / or geographies predicted as high value. The investment value(s) for the properties under consideration may then be determined based on the“closest” properties or property types and / or geographies.

CYBERSECURITY

[0163] The processes 100, 400 may also be used for cybersecurity analysis. For example, a cybersecurity company or other organization may want to perform threat (or anomalous behavior) analysis, and in particular may want explanation data associated with the threat or anomalous behavior analysis (e.g., why was a particular event, user, etc. identified as a threat or not a threat?). The computer-based reasoning model may be trained using known threats / anomalous behavior and features associated with those threats or anomalous behavior,. Data that represents neither a threat nor anomalous behavior (e.g., non-malicious access attempts, non-malicious emails, etc.) may also be used to train the computer-based reasoning model.

In some embodiments, when a new entity, user, packet, payload, routing attempt, access attempt, log file, etc. is ready for assessment, the features associated with that new entity, user, packet, payload, routing attempt, access attempt, log file, etc. may be used as input in the trained cybersecurity computer-based reasoning system. The cybersecurity computer- based reasoning system may then determine the likelihood that the entity, user, packet, payload, routing attempt, access attempt, pattern in the log file, etc. is or represents a threat or anomalous behavior. Further, explanation data, such as a conviction measures, training data used to make a decision etc., can be used to mitigate the threat or anomalous behavior and / or be provided to a human operator in order to further assess the potential threat or anomalous behavior.

[0164] Any type of cybersecurity threat or anomalous behavior can be analyzed and detected, such as denial of service (DoS), distributed DOS (DDoS), brute-force attacks (e.g., password breach attempts), compromised credentials, malware, insider threats, advanced persistent threats, phishing, spear phishing, etc. and / or anomalous traffic volume, bandwidth use, protocol use, behavior of individuals and / or accounts, logfile pattern, access or routing attempt, etc. In some embodiments the cybersecurity threat is mitigated (e.g., access is suspended, etc.) while the threat is escalated to a human operator. As a more specific example, if an email is received by the email server, the email may be provided as input to the trained cybersecurity computer-based reasoning model. The cybersecurity computer- based reasoning model may indicate that the email is a potential threat (e.g., detecting and then indicating that email includes a link to a universal resource locator that is different from the universal resource location displayed in the text of the email). In some embodiments, this email may be automatically deleted, may be quarantined, and / or flagged for review.

[0165] In some embodiments, processes 100, 400 may include determining (e.g., in response to a request) the surprisal and / or conviction of one or more data elements or aspects of the cybersecurity system. The surprisal or conviction for the one or more cybersecurity system data elements or aspects may be determined and a determination may be made whether to select or include the one or more cybersecurity system data elements or aspects in a cybersecurity system computer-based reasoning model based on the determined surprisal and / or conviction. While there are more sets of one or more cybersecurity system data elements or aspects to assess, the process may return to determine whether more cybersecurity system data elements or aspects should be included. Once there are no more cybersecurity system data elements or aspects to consider, the process can turn to controlling the cybersecurity computer-based reasoning system using the cybersecurity system computer- based reasoning model.

[0166] In some embodiments, process 100 may determine (e.g., in response to a request) synthetic data for use in the cybersecurity system computer-based reasoning model. Based on a model that uses the synthetic data, the process can cause 199 control of a cybersecurity computer-based reasoning system using process 400. For example, if the data elements are related to cybersecurity system actions, then the cybersecurity system computer-based reasoning model trained on that data will control the cybersecurity system (e.g., quarantine, delete, or flag for review, entities, data, network traffic, etc.). Process 400 proceeds by receiving 410 a cybersecurity system computer-based reasoning model. The process proceeds by receiving 420 a context. The cybersecurity system computer-based reasoning model is then used to determine 430 an action to take. The action is then performed by the control system (e.g., caused by the cybersecurity system computer-based reasoning system).

If there are more 450 contexts to consider, then the system returns to receive 410 those contexts and otherwise ceases 460. In such embodiments, the cybersecurity system computer-based reasoning model may be used to assess cybersecurity threats, etc. In some embodiments, the chosen action(s) are then performed by a control system.

EXAMPLE CONTROL HIERARCHIES [0167] In some embodiments, the technique herein may use a control hierarchy to control systems and / or cause actions to be taken (e.g., as part of causing 199 control). There are numerous example control hierarchies and many types of systems to control, and hierarchy for vehicle control is presented below. In some embodiments, only a portion of this control hierarchy is used. It is also possible to add levels to (or remove levels from) the control hierarchy.

[0168] An example control hierarchy for controlling a vehicle could be:

Primitive Layer - Active vehicle abilities (accelerate, decelerate), lateral, elevation, and orientation movements to control basic vehicle navigation

Behavior Layer - Programmed vehicle behaviors which prioritize received actions and directives and prioritize the behaviors in the action.

Unit Layer - Receives orders from command layer, issues moves/directives to the behavior layer.

Command Layers (hierarchical) - Receives orders and gives orders to

elements under its command, which may be another command layer or unit layer.

EXAMPLE DATA CASES, DATA ELEMENTS, CONTEXTS, AND OPERATIONAL SITUATIONS

[0169] In some embodiments, the cases, data cases, or data elements may include context data and action data in context-action pairs. Further, cases may relate to control of a vehicle, control of a smart voice control, health system, real estate system, image labelling systems, or any of the other examples herein. For example, context data may include data related to the operation of the vehicle, including the environment in which it is operating, and the actions taken may be of any granularity. Consider an example of data collected while a driver,

Alicia, drives around a city. The collected data could be context and action data where the actions taken can include high-level actions (e.g., drive to next intersection, exit the highway, take surface roads, etc.), mid-level actions (e.g., turn left, turn right, change lanes) and / or low-level actions (e.g., accelerate, decelerate, etc.). The contexts can include any information related to the vehicle (e.g. time until impact with closest object(s), speed, course heading, breaking distances, vehicle weight, etc.), the driver (pupillary dilation, heart rate,

attentiveness, hand position, foot position, etc.), the environment (speed limit and other local rules of the road, weather, visibility, road surface information, both transient such as moisture level as well as more permanent, such as pavement levelness, existence of potholes, etc.), traffic (congestion, time to a waypoint, time to destination, availability of alternate routes, etc.), and the like. These input data (e.g., context-action pairs for training a context-based reasoning system or input training contexts with outcome actions for training a machine learning system) can be saved and later used to help control a compatible vehicle in a compatible operational situation. The operational situation of the vehicle may include any relevant data related to the operation of the vehicle. In some embodiments, the operational situation may relate to operation of vehicles by particular individuals, in particular geographies, at particular times, and in particular conditions. For example, the operational situation may refer to a particular driver (e.g., Alicia or Carole). Alicia may be considered a cautious car driver, and Carole a faster driver. As noted above, and in particular, when approaching a stop sign, Carole may coast in and then brake at the last moment, while Alicia may slow down earlier and roll in. As another example of an operational situation, Bob may be considered the“best pilot” for a fleet of helicopters, and therefore his context and actions may be used for controlling self-flying helicopters.

[0170] In some embodiments, the operational situation may relate to the environment in which the system is operating. In the vehicle context, the locale may be a geographic area of any size or type and may be determined by systems that utilize machine learning. For example, an operational situation may be“highway driving” while another is“side street driving”. An operational situation may be related to an area, neighborhood, city, region, state, country, etc. For example, one operational situation may relate to driving in Raleigh, NC and another may be driving in Pittsburgh, PA. An operational situation may relate to safe or legal driving speeds. For example, one operational situation may be related to roads with forty-five miles per hour speed limits, and another may relate to turns with a recommended speed of 20 miles per hour. The operational situation may also include aspects of the environment such as road congestion, weather or road conditions, time of day, etc. The operational situation may also include passenger information, such as whether to hurry (e.g., drive faster), whether to drive smoothly, technique for approaching stop signs, red lights, other objects, what relative velocity to take turns, etc. The operational situation may also include cargo information, such as weight, hazardousness, value, fragility of the cargo, temperature sensitivity, handling instructions, etc.

[0171] In some embodiments, the context and action may include system maintenance information. In the vehicle context, the context may include information for timing and / or wear-related information for individual or sets of components. For example, the context may include information on the timing and distance since the last change of each fluid, each belt, each tire (and possibly when each was rotated), the electrical system, interior and exterior materials (such as exterior paint, interior cushions, passenger entertainment systems, etc.), communication systems, sensors (such as speed sensors, tire pressure monitors, fuel gauges, compasses, global positioning systems (GPS), RADARs, LiDARs, cameras, barometers, thermal sensors, accelerometers, strain gauges, noise/sound measurement systems, etc.), the engine(s), structural components of the vehicle (wings, blades, struts, shocks, frame, hull, etc.), and the like. The action taken may include inspection, preventative maintenance, and / or a failure of any of these components. As discussed elsewhere herein, having context and actions related to maintenance may allow the techniques to predict when issues will occur with future vehicles and / or suggest maintenance. For example, the context of an automobile may include the distance traveled since the timing belt was last replaced. The action associated with the context may include inspection, preventative replacement, and / or failure of the timing belt. Further, as described elsewhere herein, the contexts and actions may be collected for multiple operators and / or vehicles. As such, the timing of inspection, preventative maintenance and / or failure for multiple automobiles may be determined and later used for predictions and messaging.

[0172] Causing performance of an identified action can include causing a control system to control the target system based on the identified action. In the self-controlled vehicle context, this may include sending a signal to a real car, to a simulator of a car, to a system or device in communication with either, etc. Further, the action to be caused can be simulated / predicted without showing graphics, etc. For example, the techniques might cause performance of actions in the manner that includes, determining what action would be take, and determining whether that result would be anomalous, and performing the techniques herein based on the determination that such state would be anomalous based on that determination, all without actually generating the graphics and other characteristics needed for displaying the results needed in a graphical simulator (e.g., a graphical simulator might be similar to a computer game).

HARDWARE OVERVIEW

[0173] According to some embodiments, the techniques described herein are

implemented by one or more special-purpose computing devices. The special-purpose computing devices may be hard-wired to perform the techniques, or may include digital electronic devices such as one or more application-specific integrated circuits (ASICs) or field programmable gate arrays (FPGAs) that are persistently programmed to perform the techniques, or may include one or more general purpose hardware processors programmed to perform the techniques pursuant to program instructions in firmware, memory, other storage, or a combination. Such special-purpose computing devices may also combine custom hard wired logic, ASICs, or FPGAs with custom programming to accomplish the techniques. The special-purpose computing devices may be desktop computer systems, portable computer systems, handheld devices, networking devices or any other device that incorporates hard wired and/or program logic to implement the techniques.

[0174] For example, FIG. 3 is a block diagram that illustrates a computer system 300 upon which an embodiment of the invention may be implemented. Computer system 300 includes a bus 302 or other communication mechanism for communicating information, and a hardware processor 304 coupled with bus 302 for processing information. Hardware processor 304 may be, for example, a general purpose microprocessor.

[0175] Computer system 300 also includes a main memory 306, such as a random access memory (RAM) or other dynamic storage device, coupled to bus 302 for storing information and instructions to be executed by processor 304. Main memory 306 also may be used for storing temporary variables or other intermediate information during execution of instructions to be executed by processor 304. Such instructions, when stored in non-transitory storage media accessible to processor 304, render computer system 300 into a special-purpose machine that is customized to perform the operations specified in the instructions.

[0176] Computer system 300 further includes a read only memory (ROM) 308 or other static storage device coupled to bus 302 for storing static information and instructions for processor 304. A storage device 310, such as a magnetic disk, optical disk, or solid-state drive is provided and coupled to bus 302 for storing information and instructions.

[0177] Computer system 300 may be coupled via bus 302 to a display 312, such as an OLED, LED or cathode ray tube (CRT), for displaying information to a computer user. An input device 314, including alphanumeric and other keys, is coupled to bus 302 for communicating information and command selections to processor 304. Another type of user input device is cursor control 316, such as a mouse, a trackball, or cursor direction keys for communicating direction information and command selections to processor 304 and for controlling cursor movement on display 312. This input device typically has two degrees of freedom in two axes, a first axis (e.g., x) and a second axis (e.g., y), that allows the device to specify positions in a plane. The input device 314 may also have multiple input modalities, such as multiple 2-axes controllers, and / or input buttons or keyboard. This allows a user to input along more than two dimensions simultaneously and / or control the input of more than one type of action.

[0178] Computer system 300 may implement the techniques described herein using customized hard-wired logic, one or more ASICs or FPGAs, firmware and/or program logic which in combination with the computer system causes or programs computer system 300 to be a special-purpose machine. According to some embodiments, the techniques herein are performed by computer system 300 in response to processor 304 executing one or more sequences of one or more instructions contained in main memory 306. Such instructions may be read into main memory 306 from another storage medium, such as storage device 310. Execution of the sequences of instructions contained in main memory 306 causes processor 304 to perform the process steps described herein. In alternative embodiments, hard-wired circuitry may be used in place of or in combination with software instructions.

[0179] The term“storage media” as used herein refers to any non-transitory media that store data and/or instructions that cause a machine to operate in a specific fashion. Such storage media may comprise non-volatile media and/or volatile media. Non-volatile media includes, for example, optical disks, magnetic disks, or solid-state drives, such as storage device 310. Volatile media includes dynamic memory, such as main memory 306. Common forms of storage media include, for example, a floppy disk, a flexible disk, hard disk, solid- state drive, magnetic tape, or any other magnetic data storage medium, a CD-ROM, any other optical data storage medium, any physical medium with patterns of holes, a RAM, a PROM, and EPROM, a FLASH-EPROM, NVRAM, any other memory chip or cartridge.

[0180] Storage media is distinct from but may be used in conjunction with transmission media. Transmission media participates in transferring information between storage media. For example, transmission media includes coaxial cables, copper wire and fiber optics, including the wires that comprise bus 302. Transmission media can also take the form of acoustic or light waves, such as those generated during radio-wave and infra-red data communications.

[0181] Various forms of media may be involved in carrying one or more sequences of one or more instructions to processor 304 for execution. For example, the instructions may initially be carried on a magnetic disk or solid-state drive of a remote computer. The remote computer can load the instructions into its dynamic memory and send the instructions over a telephone line using a modem. A modem local to computer system 300 can receive the data on the telephone line and use an infra-red transmitter to convert the data to an infra-red signal. An infra-red detector can receive the data carried in the infra-red signal and appropriate circuitry can place the data on bus 302. Bus 302 carries the data to main memory 306, from which processor 304 retrieves and executes the instructions. The instructions received by main memory 306 may optionally be stored on storage device 310 either before or after execution by processor 304.

[0182] Computer system 300 also includes a communication interface 318 coupled to bus 302. Communication interface 318 provides a two-way data communication coupling to a network link 320 that is connected to a local network 322. For example, communication interface 318 may be an integrated services digital network (ISDN) card, cable modem, satellite modem, or a modem to provide a data communication connection to a corresponding type of telephone line. As another example, communication interface 318 may be a local area network (LAN) card to provide a data communication connection to a compatible LAN. Wireless links may also be implemented. In any such implementation, communication interface 318 sends and receives electrical, electromagnetic or optical signals that carry digital data streams representing various types of information. Such a wireless link could be a Bluetooth, Bluetooth Low Energy (BLE), 802.11 WiFi connection, or the like.

[0183] Network link 320 typically provides data communication through one or more networks to other data devices. For example, network link 320 may provide a connection through local network 322 to a host computer 324 or to data equipment operated by an Internet Service Provider (ISP) 326. ISP 326 in turn provides data communication services through the world wide packet data communication network now commonly referred to as the“Internet” 328. Local network 322 and Internet 328 both use electrical, electromagnetic or optical signals that carry digital data streams. The signals through the various networks and the signals on network link 320 and through communication interface 318, which carry the digital data to and from computer system 300, are example forms of transmission media.

[0184] Computer system 300 can send messages and receive data, including program code, through the network(s), network link 320 and communication interface 318. In the Internet example, a server 330 might transmit a requested code for an application program through Internet 328, ISP 326, local network 322 and communication interface 318.

[0185] The received code may be executed by processor 304 as it is received, and/or stored in storage device 310, or other non-volatile storage for later execution.

[0186] In the foregoing specification, embodiments of the invention have been described with reference to numerous specific details that may vary from implementation to

implementation. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense. The sole and exclusive indicator of the scope of the invention, and what is intended by the applicants to be the scope of the invention, is the literal and equivalent scope of the set of claims that issue from this application, in the specific form in which such claims issue, including any subsequent correction.