Support vector machines resilient against training data integrity attacks. (December 2019)

• Record Type: Journal Article
• Authors: Weerasinghe, Sandamal; Erfani, Sarah M.; Alpcan, Tansu; Leckie, Christopher
• Is Part Of: Pattern recognition. Volume 96(2019:Dec.)
• Publication Date: 2019-12
• Subjects: Support Vector Machines -- Integrity attack; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2019.106985
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: Support Vector Machines are designed to withstand noise in data. But they are vulnerable to integrity attacks by adversaries. Projecting data to lower dimensional spaces in specific directions may reduce the adversary's effects. Game theory can be used to predict the adversary's actions and take proactive precautions. Abstract: Support Vector Machines (SVMs) are vulnerable to integrity attacks, where malicious attackers distort the training data in order to compromise the decision boundary of the learned model. With increasing real-world applications of SVMs, malicious data that is classified as innocuous may have harmful consequences. This paper presents a novel framework that utilizes adversarial learning, nonlinear data projections, and game theory to improve the resilience of SVMs against such training-data-integrity attacks. The proposed approach introduces a layer of uncertainty through the use of random projections on top of the learners, making it challenging for the adversary to guess the specific configurations of the learners. To find appropriate projection directions, we introduce novel indices that ensure the contraction of the data and maximize the detection accuracy. Experiments with benchmark data sets show increases in detection rates up to 13.5% for OCSVMs and up to 14.1% for binary SVMs under different attack algorithms when compared with the respective base algorithms.