Runtime safety filters for learned manipulation policies typically define unsafe states as unions of object-wise keep-out regions. This representation can be unnecessarily restrictive for hazards that depend on a joint spatial relation, such as battery recycling, where a conductive payload can short a charged cell only...
Yu-Xin Cao, Wei Song, Xiang-Lin Yang et al.· 0 citations
Adversarial patches to Vision-Language-Action (VLA) policies can cause both immediate action corruption and persistent state effects that remain after the patch is removed. Existing evaluations largely focus on continuous attacks and do not separate these two effects. We introduce a state-restoration protocol that remo...
Enjia Wu, Fu-Sen Guo, Yu-Xin Cao et al.· 0 citations
Implemented in vLLM and SGLang 0.5.20 across three cache paths, provenance binding eliminates unsafe reuse while preserving legitimate sharing and retention grows with provenance diversity.
Wei Song, Yu-Xin Cao, Xiao-Dong Zheng et al.· 0 citations
When asked which of two events came first, video large language models can fail in two opposite ways: cave to a false claim, or reject a true one. Prior video sycophancy work measures only the first and mitigates it by teaching the model to trust the user less, a fix known in text and image models to worsen the second....
Yu-Xin Cao, Wei Song, Jing-Ling Xue et al.· 0 citations
Video watermarking underpins copyright protection and provenance for generated media, yet almost every video is compressed by a codec before it is stored or shared. A codec discards precisely the perceptually redundant components that most watermarks rely on, so the payload is often lost even when the marked video look...
Yu-Xin Cao, Hao Yang, Zi-Qi Ding et al.· 1 citation
KeyBound is presented, a learned audio watermark that restores the two ingredients classical watermarking supplied and learned schemes set aside, a secret key and a host-aware carrier, so the key governs payload access while the host-conditioned carrier resists direct transplantation.
Bang-Shuo Zhu, Yu-Xin Cao, Wei-Fei Jin et al.· 0 citations
DefTEval is introduced, a controlled evaluation framework that systematically assesses whether input-level adversarial defenses, which operate on the pixel content of already-sampled frames, can mitigate observation-level attacks and demonstrates that securing VideoLLMs requires system-level robustness mechanisms spann...
Bang-Shuo Zhu, Wei Song, Yu-Xin Cao et al.· 0 citations
SpreadMark keeps the embedded watermark imperceptible, maintaining high perceptual quality on both COCO and DIV2K, and is the only evaluated method retaining high detection under both the regeneration and the latent-space sparsification settings the authors test.
Wei Song, Yu-Xin Cao, Zhen-Chang Xing et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
