Purifying Backdoored Large Vision-Language Models by Removing Hijacked Directions
OrthoPurify is proposed, a more efficient method to purify backdoored model weights via one-step orthogonal projection, which reduces the attack success rate to near zero while preserving the original performance across diverse benchmarks, without retraining the backdoored model or introducing inference-time overhead.
Bo Yang, Hao-Chen Zhou, Zhi-Fang Zhang et al.
· 0 citations