CANARY: An Adversarial Robustness Evaluation Platform for Deep Learning Models on Image Classification
The vulnerability of deep-learning-based image classification models to erroneous conclusions in the presence of small perturbations crafted by attackers has prompted attention to the question of the models’ robustness level. However, the question of how to comprehensively and fairly measure the adv...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-08-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/12/17/3665 |
| _version_ | 1797582683527708672 |
|---|---|
| author | Jiazheng Sun Li Chen Chenxiao Xia Da Zhang Rong Huang Zhi Qiu Wenqi Xiong Jun Zheng Yu-An Tan |
| author_facet | Jiazheng Sun Li Chen Chenxiao Xia Da Zhang Rong Huang Zhi Qiu Wenqi Xiong Jun Zheng Yu-An Tan |
| author_sort | Jiazheng Sun |
| collection | DOAJ |
| description | The vulnerability of deep-learning-based image classification models to erroneous conclusions in the presence of small perturbations crafted by attackers has prompted attention to the question of the models’ robustness level. However, the question of how to comprehensively and fairly measure the adversarial robustness of models with different structures and defenses as well as the performance of different attack methods has never been accurately answered. In this work, we present the design, implementation, and evaluation of Canary, a platform that aims to answer this question. Canary uses a common scoring framework that includes 4 dimensions with 26 (sub)metrics for evaluation. First, Canary generates and selects valid adversarial examples and collects metrics data through a series of tests. Then it uses a two-way evaluation strategy to guide the data organization and finally integrates all the data to give the scores for model robustness and attack effectiveness. In this process, we use Item Response Theory (IRT) for the first time to ensure that all the metrics can be fairly calculated into a score that can visually measure the capability. In order to fully demonstrate the effectiveness of Canary, we conducted large-scale testing of 15 representative models trained on the ImageNet dataset using 12 white-box attacks and 12 black-box attacks and came up with a series of in-depth and interesting findings. This further illustrates the capabilities and strengths of Canary as a benchmarking platform. Our paper provides an open-source framework for model robustness evaluation, allowing researchers to perform comprehensive and rapid evaluations of models or attack/defense algorithms, thus inspiring further improvements and greatly benefiting future work. |
| first_indexed | 2024-03-10T23:24:54Z |
| format | Article |
| id | doaj.art-a20c0eb100d947509e507917121a6006 |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-10T23:24:54Z |
| publishDate | 2023-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-a20c0eb100d947509e507917121a60062023-11-19T08:02:20ZengMDPI AGElectronics2079-92922023-08-011217366510.3390/electronics12173665CANARY: An Adversarial Robustness Evaluation Platform for Deep Learning Models on Image ClassificationJiazheng Sun0Li Chen1Chenxiao Xia2Da Zhang3Rong Huang4Zhi Qiu5Wenqi Xiong6Jun Zheng7Yu-An Tan8School of Cyberspace Science & Technology, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Computer Science & Technology, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Computer Science & Technology, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Cyberspace Science & Technology, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Cyberspace Science & Technology, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Cyberspace Science & Technology, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Cyberspace Science & Technology, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Cyberspace Science & Technology, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Cyberspace Science & Technology, Beijing Institute of Technology, Beijing 100081, ChinaThe vulnerability of deep-learning-based image classification models to erroneous conclusions in the presence of small perturbations crafted by attackers has prompted attention to the question of the models’ robustness level. However, the question of how to comprehensively and fairly measure the adversarial robustness of models with different structures and defenses as well as the performance of different attack methods has never been accurately answered. In this work, we present the design, implementation, and evaluation of Canary, a platform that aims to answer this question. Canary uses a common scoring framework that includes 4 dimensions with 26 (sub)metrics for evaluation. First, Canary generates and selects valid adversarial examples and collects metrics data through a series of tests. Then it uses a two-way evaluation strategy to guide the data organization and finally integrates all the data to give the scores for model robustness and attack effectiveness. In this process, we use Item Response Theory (IRT) for the first time to ensure that all the metrics can be fairly calculated into a score that can visually measure the capability. In order to fully demonstrate the effectiveness of Canary, we conducted large-scale testing of 15 representative models trained on the ImageNet dataset using 12 white-box attacks and 12 black-box attacks and came up with a series of in-depth and interesting findings. This further illustrates the capabilities and strengths of Canary as a benchmarking platform. Our paper provides an open-source framework for model robustness evaluation, allowing researchers to perform comprehensive and rapid evaluations of models or attack/defense algorithms, thus inspiring further improvements and greatly benefiting future work.https://www.mdpi.com/2079-9292/12/17/3665AI securityadversarial robustness evaluationadversarial attackdeep model |
| spellingShingle | Jiazheng Sun Li Chen Chenxiao Xia Da Zhang Rong Huang Zhi Qiu Wenqi Xiong Jun Zheng Yu-An Tan CANARY: An Adversarial Robustness Evaluation Platform for Deep Learning Models on Image Classification Electronics AI security adversarial robustness evaluation adversarial attack deep model |
| title | CANARY: An Adversarial Robustness Evaluation Platform for Deep Learning Models on Image Classification |
| title_full | CANARY: An Adversarial Robustness Evaluation Platform for Deep Learning Models on Image Classification |
| title_fullStr | CANARY: An Adversarial Robustness Evaluation Platform for Deep Learning Models on Image Classification |
| title_full_unstemmed | CANARY: An Adversarial Robustness Evaluation Platform for Deep Learning Models on Image Classification |
| title_short | CANARY: An Adversarial Robustness Evaluation Platform for Deep Learning Models on Image Classification |
| title_sort | canary an adversarial robustness evaluation platform for deep learning models on image classification |
| topic | AI security adversarial robustness evaluation adversarial attack deep model |
| url | https://www.mdpi.com/2079-9292/12/17/3665 |
| work_keys_str_mv | AT jiazhengsun canaryanadversarialrobustnessevaluationplatformfordeeplearningmodelsonimageclassification AT lichen canaryanadversarialrobustnessevaluationplatformfordeeplearningmodelsonimageclassification AT chenxiaoxia canaryanadversarialrobustnessevaluationplatformfordeeplearningmodelsonimageclassification AT dazhang canaryanadversarialrobustnessevaluationplatformfordeeplearningmodelsonimageclassification AT ronghuang canaryanadversarialrobustnessevaluationplatformfordeeplearningmodelsonimageclassification AT zhiqiu canaryanadversarialrobustnessevaluationplatformfordeeplearningmodelsonimageclassification AT wenqixiong canaryanadversarialrobustnessevaluationplatformfordeeplearningmodelsonimageclassification AT junzheng canaryanadversarialrobustnessevaluationplatformfordeeplearningmodelsonimageclassification AT yuantan canaryanadversarialrobustnessevaluationplatformfordeeplearningmodelsonimageclassification |