TY  - RPRT
TI  - Purifying Backdoored Large Vision-Language Models by Removing Hijacked Directions
AU  - Bojun Yang
AU  - Haochen Zhou
AU  - Zhifang Zhang
AU  - Haobo Wang
AU  - Songze Li
AU  - Lei Feng
PY  - 2026
UR  - https://arxiv.org/abs/2610.09941
ID  - 2610.09941
ER  -