Pith. sign in

Backdoor Mitigation by Distance-Driven Detoxification

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it
abstract

Backdoor attacks undermine the integrity of machine learning models by allowing attackers to manipulate predictions using poisoned training data. Such attacks lead to targeted misclassification when specific triggers are present, while the model behaves normally under other conditions. This paper considers a post-training backdoor defense task, aiming to detoxify the backdoors in pre-trained models. We begin by analyzing the underlying issues of vanilla fine-tuning and observe that it is often trapped in regions with low loss for both clean and poisoned samples. Motivated by such observations, we propose Distance-Driven Detoxification (D3), an innovative approach that reformulates backdoor defense as a constrained optimization problem. Specifically, D3 promotes the model's departure from the vicinity of its initial weights, effectively reducing the influence of backdoors. Extensive experiments on state-of-the-art (SOTA) backdoor attacks across various model architectures and datasets demonstrate that D3 not only matches but often surpasses the performance of existing SOTA post-training defense techniques.

citation-role summary

background 1

citation-polarity summary

fields

cs.CR 1

years

2025 1

verdicts

CONDITIONAL 1

roles

background 1

polarities

unclear 1

representative citing papers

Revisiting the Auxiliary Data in Backdoor Purification

cs.CR · 2025-02-11 · conditional · novelty 6.0

Guided Input Calibration aligns any auxiliary dataset with a victim model's learned features before backdoor purification, consistently improving clean accuracy across dataset types with variable effects on attack success rate.

citing papers explorer

Showing 1 of 1 citing paper.

  • Revisiting the Auxiliary Data in Backdoor Purification cs.CR · 2025-02-11 · conditional · none · ref 35 · internal anchor

    Guided Input Calibration aligns any auxiliary dataset with a victim model's learned features before backdoor purification, consistently improving clean accuracy across dataset types with variable effects on attack success rate.