Title:
INFORMATION PROCESSING METHOD AND INFORMATION PROCESSING SYSTEM
Document Type and Number:
WIPO Patent Application WO/2023/002648
Kind Code:
A1
Abstract:
The present invention presents the result of evaluating training data used for training a machine learning model. This information processing method for executing processing relating to training data used for training a machine learning model comprises a determination step of determining the characteristic of each piece of training data on the basis of the result of inference by the machine learning model for the training data, and a presentation step of presenting the result of evaluating the training data on the basis of the determined characteristic. The determination step involves determining a physical characteristic such as the mass, size, or acting force including attraction and repulsion of an object corresponding to the training data on the basis of an expected value of each label output by the machine learning model for the training data.
Inventors:
TSUJIKAWA TOMONOBU (JP)
OOBUCHI YUKIO (JP)
KOBAYASHI YOSHIYUKI (JP)
OOBUCHI YUKIO (JP)
KOBAYASHI YOSHIYUKI (JP)
Application Number:
PCT/JP2022/003897
Publication Date:
January 26, 2023
Filing Date:
February 01, 2022
Export Citation:
Assignee:
SONY GROUP CORP (JP)
International Classes:
G06N20/00; G16Z99/00
Foreign References:
| JP2021060692A | 2021-04-15 | |||
| JP2020140226A | 2020-09-03 | |||
| JP2020197875A | 2020-12-10 | |||
| JP2018194919A | 2018-12-06 |
Other References:
GUMELAR AGUSTINUS BIMO: "An Anatomy of Machine Learning Data Visualization", 2019 INTERNATIONAL SEMINAR ON APPLICATION FOR TECHNOLOGY OF INFORMATION AND COMMUNICATION (ISEMANTIC), 21 September 2019 (2019-09-21), pages 1 - 6, XP033651025, DOI: 10.1109/ISEMANTIC.2019.8884340
GRAD-CAM: VISUAL EXPLANATIONS FROM DEEP NETWORKS VIA GRADIENT-BASED LOCALIZATION, Retrieved from the Internet
"Why Should I Trust You?", EXPLAINING THE PREDICTIONS OF ANY CLASSIFIER, Retrieved from the Internet
INTERPRETABILITY BEYOND FEATURE ATTRIBUTION: QUANTITATIVE TESTING WITH CONCEPT ACTIVATION VECTORS (TCAV, Retrieved from the Internet
GRAD-CAM: VISUAL EXPLANATIONS FROM DEEP NETWORKS VIA GRADIENT-BASED LOCALIZATION, Retrieved from the Internet
"Why Should I Trust You?", EXPLAINING THE PREDICTIONS OF ANY CLASSIFIER, Retrieved from the Internet
INTERPRETABILITY BEYOND FEATURE ATTRIBUTION: QUANTITATIVE TESTING WITH CONCEPT ACTIVATION VECTORS (TCAV, Retrieved from the Internet
Attorney, Agent or Firm:
MIYATA, Masaaki et al. (JP)
Download PDF:
Previous Patent: BUSINESS ASSISTANCE SYSTEM AND BUSINESS ASSISTANCE METHOD
Next Patent: CLEANING DEVICE FOR CONTAINER
Next Patent: CLEANING DEVICE FOR CONTAINER