Title:
LEARNING DEVICE, CLASSIFICATION DEVICE, LEARNING METHOD, AND LEARNING PROGRAM
Document Type and Number:
WIPO Patent Application WO/2021/070394
Kind Code:
A1
Abstract:
A score calculation unit (123) calculates, by using a score function to calculate a score from a feature of data in accordance with a parameter, a score for each of one or more pieces of data which are known as a negative instance or a positive instance. Further, in a classification result in the case where classification is performed on the basis of the score calculated by the score calculation unit (123), an index calculation unit (124) calculates an index which becomes larger as a true positive rate in the case where a false positive rate is within a predetermined section is increased, and becomes larger as a proportion of positive instance data out of data having a score equal to a predetermined value is increased. Further, an update unit (125) updates the parameter in such a manner that the index calculated by the index calculation unit (124) is optimized.
More Like This:
Inventors:
NISHIYAMA TAISHI (JP)
KUMAGAI ATSUTOSHI (JP)
KAMIYA KAZUNORI (JP)
KUMAGAI ATSUTOSHI (JP)
KAMIYA KAZUNORI (JP)
Application Number:
PCT/JP2019/040337
Publication Date:
April 15, 2021
Filing Date:
October 11, 2019
Export Citation:
Assignee:
NIPPON TELEGRAPH & TELEPHONE (JP)
International Classes:
G06N20/00
Foreign References:
| JP2017102540A | 2017-06-08 |
Other References:
KUWABARA, AKIYUKI: "A Perceptron Learning Rule for Locally Improving the ROC Curve", IEICE TECH. REP., vol. 111, no. 483, 7 March 2012 (2012-03-07), pages 399 - 404, ISSN: 0913-5685
Attorney, Agent or Firm:
SAKAI INTERNATIONAL PATENT OFFICE (JP)
Download PDF: