Title:
LEARNER ESTIMATING DEVICE, LEARNER ESTIMATION METHOD, RISK EVALUATION DEVICE, RISK EVALUATION METHOD, AND PROGRAM
Document Type and Number:
WIPO Patent Application WO/2020/075462
Kind Code:
A1
Abstract:
The present invention effectively estimates a learner for classification. This learner estimating device assumes, as an attack target, a learner for classification tasks that outputs the type of inputted observation data as label data, the learner estimating device comprising a recording unit, an inquiry unit, a capturing unit, and a learning unit. The recording unit keeps a record of a plurality of predetermined observation data. The inquiry unit inquires of the learner that is the attack target for each of the observation data recorded in the recording unit, acquires label data, and records the acquired label data in the recording unit in association with the observation data. The capturing unit inputs the observation data recorded in the recording unit and the label data associated with the observation data to the learning unit. The learning unit is characterized by the use of an activation function that outputs an ambiguous value predetermined in a process for establishing the predictive result of classification, the learning unit learning through use of the inputted observation data and label data.
Inventors:
OKADA RINA (JP)
HASEGAWA SATOSHI (JP)
HASEGAWA SATOSHI (JP)
Application Number:
PCT/JP2019/036595
Publication Date:
April 16, 2020
Filing Date:
September 18, 2019
Export Citation:
Assignee:
NIPPON TELEGRAPH & TELEPHONE (JP)
International Classes:
G06N3/04; G06F21/57
Other References:
SHOKRI, REZA ET AL.: "Membership Inference Attacks Against Machine Learning Models", PROCEEDINGS OF 2017 IEEE SYMPOSIUM ON SECURITY AND PRIVACY, 26 May 2017 (2017-05-26), pages 3 - 18, XP033108084, Retrieved from the Internet [retrieved on 20191118], DOI: 10.1109/SP.2017.41
Attorney, Agent or Firm:
NAKAO, Naoki et al. (JP)
Download PDF: