Title:
DECISION LIST LEARNING DEVICE, DECISION LIST LEARNING METHOD, AND DECISION LIST LEARNING PROGRAM
Document Type and Number:
WIPO Patent Application WO/2020/059136
Kind Code:
A1
Abstract:
An input unit 81 accepts a set of rules including a condition and a prediction, as well as observation data and a correct answer pair. A probabilistic decision list generation unit 82 allocates each rule included in the set of rules to a plurality of positions in a decision list, with an occurrence degree indicating the degree of occurrence being attached to each rule. A learning unit 83 updates a parameter that determines the occurrence degree so as to reduce a difference between the correct answer and an integrated prediction obtained by integrating, on the basis of the occurrence degree, the prediction of a rule with which the observation data satisfies the condition.
Inventors:
OKAJIMA YUZURU (JP)
SADAMASA KUNIHIKO (JP)
SADAMASA KUNIHIKO (JP)
Application Number:
PCT/JP2018/035162
Publication Date:
March 26, 2020
Filing Date:
September 21, 2018
Export Citation:
Assignee:
NEC CORP (JP)
International Classes:
G06N99/00
Domestic Patent References:
| WO2017168460A1 | 2017-10-05 |
Foreign References:
| JPH07175770A | 1995-07-14 | |||
| JPH1196010A | 1999-04-09 |
Other References:
SHIRAKI, NOBUYUKI ET AL.: "Bunsetsu Identification with Sequential Use of Plural Decision Lists", JOURNAL OF NATURAL LANGUAGE PROCESSING, vol. 7, no. 4, pages 232 - 233
Attorney, Agent or Firm:
IWAKABE Fuyuki et al. (JP)
Download PDF: