In this work, we propose a new and general framework to defend against backdoor attacks, inspired by the fact that attack triggers usually follow a \textsc{ specific} type of attacking pattern, and therefore, poisoned training examples have greater impacts on each other during training. We introduce the notion of the { \it influence graph}, which consists of nodes and edges respectively representative of individual training points and associated pair-wise influences. The influence between a pair of training points represents the impact of removing one training point on the prediction of another, approximated by the influence function \citep{ koh2017understanding}. Malicious training points are extracted by finding the maximum average sub-graph subject to a particular size. Extensive experiments on computer vision and natural language processing tasks demonstrate the effectiveness and generality of the proposed framework.

Categories: cs.LG cs.AI cs.CL cs.CV

A General Framework for Defending Against Backdoor Attacks via Influence Graph