Title:
IMAGE MODEL TESTING METHOD AND APPARATUS, ELECTRONIC DEVICE AND STORAGE MEDIUM
Document Type and Number:
WIPO Patent Application WO/2021/056746
Kind Code:
A1
Abstract:
Provided is an image model testing method, the method comprising: acquiring an original image sample; inputting the original image sample into a trained mainstream image classification model; using the mainstream image classification model and a momentum-based gradient iteration algorithm to perform an adversarial attack on the original image sample to obtain an adversarial image; acquiring a first recognition result obtained after an image model to be tested recognizes the original image sample, and acquiring a second recognition result obtained after the image model to be tested recognizes the adversarial image; determining whether the first recognition result is consistent with the second recognition result; and if the first recognition result is consistent with the second recognition result, determining that the image model to be tested successfully recognizes the adversarial image. Further provided in the present application are an image model testing apparatus, an electronic device and a storage medium. According to the present application, the security of a deep neural network model can be tested.
Inventors:
WANG JIANZONG (CN)
HUANG ZHANGCHENG (CN)
HUANG ZHANGCHENG (CN)
Application Number:
PCT/CN2019/118027
Publication Date:
April 01, 2021
Filing Date:
November 13, 2019
Export Citation:
Assignee:
PING AN TECH SHENZHEN CO LTD (CN)
International Classes:
G06K9/62; G06F21/57
Foreign References:
| CN110222831A | 2019-09-10 | |||
| CN109492582A | 2019-03-19 | |||
| CN109165671A | 2019-01-08 | |||
| CN110245598A | 2019-09-17 | |||
| CN108615048A | 2018-10-02 | |||
| CN108537271A | 2018-09-14 | |||
| US20170365038A1 | 2017-12-21 | |||
| US20180225823A1 | 2018-08-09 |
Other References:
LUO, JIA ET AL.: "Generative adversarial network: An overview", CHINESE JOURNAL OF SCIENTIFIC INSTRUMENT, vol. 40, no. 3, 31 March 2019 (2019-03-31), pages 74 - 84, XP055795533
Attorney, Agent or Firm:
SHENZHEN SCIENBIZIP INTELLECTUAL PROPERTY AGENCY CO., LTD. (CN)
Download PDF: