ACHIEVING ADVERSARIAL ROBUSTNESS REQUIRES AN ACTIVE TEACHER

• 作者: Chao Ma Lexing Ying
• 来源期刊: 计算数学（英文版）
• 年，卷（期）: 2021,（6）
• 页码范围: 880-896
• 页数: 17页
• 语种: 英文
• DOI: 10.4208/jcm.2105-m2020-0310

摘要

A new understanding of adversarial examples and adversarial robustness is proposed by decoupling the data generator and the label generator (which we call the teacher).In our framework,adversarial robustness is a conditional concept--the student model is not absolutely robust,but robust with respect to the teacher.Based on the new understand-ing,we claim that adversarial examples exist because the student cannot obtain sufficient information of the teacher from the training data.Various ways of achieving robustness is compared.Theoretical and numerical evidence shows that to efficiently attain robustness,a teacher that actively provides its information to the student may be necessary.