← all papers · overview

Backdoor Vectors: A Task Arithmetic View On Backdoor Attacks And Defenses

Abstract

Model merging (MM) recently emerged as an effective method for combining large deep learning models. However, it poses significant security risks. Recent research shows that it is highly susceptible to backdoor attacks, which introduce a hidden trigger into a single fine-tuned model instance that allows the adversary to control the output of the final merged model at inference time. In this work, we propose a simple framework for understanding backdoor attacks by treating the attack itself as a task vector. is calculated as the difference between the weights of a fine-tuned backdoored model and fine-tuned clean model. BVs reveal new insights into attacks understanding and a more effective framework to measure their similarity and transferability. Furthermore, we propose a novel method that enhances backdoor resilience through merging dubbed that combines multiple attacks into a single one. We identify the core vulnerability

Related papers

Ranked by semantic similarity — how closely each paper's abstract matches this one (100% = near-identical topic).