TY  - RPRT
TI  - One-to-More: High-Fidelity Training-Free Anomaly Generation with Attention Control
AU  - Haoxiang Rao
AU  - Zhao Wang
AU  - Chenyang Si
AU  - Yan Lyu
AU  - Yuanyi Duan
AU  - Fang Zhao
AU  - Caifeng Shan
PY  - 2026
UR  - https://arxiv.org/abs/2603.18093
ID  - 2603.18093
ER  -