Defending Wearable VLMs Against Private Attribute Inference
Token-Guided Attribute Privacy (TGAP), a pre-LLM token disentangler that learns a residual transformation of visual tokens before they leave the trusted boundary, is proposed, suggesting that securing the compact token interface is a practical path toward privacy-preserving wearable multimodal AI.
Zhi-Min Li, Pan Wang, Jing-Xiang Chen et al.
· 0 citations