14 papers
SuperAI: Merging Capabilities Across Heterogeneous Foundation Models
Jingtong Dou, Chuancheng Shi, Wenhua Wu +4
Enabling the reuse and composition of knowledge and capabilities distributed across foundation models is an important step toward building scalable artificial intelligence (AI) sys…
A note on stability conditions on product-quotient surfaces
Uttaran Dutta
Let be a product-quotient surface, which is by definition a minimal resolution of , where are smooth curves of genus and is a finite group…
Hub-Spectral Activation of Latent Multimodal Knowledge
Ying Guo, Haidong Chen, Linrui Xu +7
Multimodal representation learning seeks shared representations for cross-modal retrieval and knowledge transfer. Hub-based binding reduces pairwise supervision costs, but separate…
V-FIND: Revealing the Intrinsic Forgery Knowledge Encoded in Video Forgery Detectors
Shichao Kan, Chengpeng Hong, Jingtong Dou +8
As generated videos become increasingly realistic, reliable video forgery detection is increasingly important. Existing studies typically optimize and use video forgery detectors a…
Moving the Safety Barrier: Dynamic Routing Adaptive Alignment Against White-Box Attacks
Shangze Li, Chuancheng Shi, Simiao Xie +6
With the widespread deployment of large foundation models (LFMs) in open environments, safety threats are shifting from black-box jailbreaks toward white-box attacks that directly…
No Single Neuron of Failure: Distributed Safety Alignment Against White-Box Attacks
Simiao Xie, Chuancheng Shi, Shangze Li +5
With the rapid release of open-weight large foundation models, safety threats are shifting from black-box jailbreaks to neuron-level white-box attacks that directly identify and ma…
Decodable Is Not Grounded: A Vision-Ablation Arbiter for VLM Spatial Reasoning
Chih-Ting Liao, Fei Shen, Xin Cao +1
The standard way to read latent knowledge out of a model, a linear probe confirmed by a steering recovery, can systematically overstate what a vision-language model (VLM) actually…
Latent Anomaly Knowledge Excavation: Unveiling Sparse Sensitive Neurons in Vision-Language Models
Shaotian Li, Shangze Li, Chuancheng Shi +5
Large-scale vision-language models (VLMs) exhibit remarkable zero-shot capabilities, yet the internal mechanisms driving their anomaly detection (AD) performance remain poorly unde…
Towards Knowledgeable Deep Research: Framework and Benchmark
Wenxuan Liu, Zixuan Li, Long Bai +13
Deep Research (DR) requires LLM agents to autonomously perform multi-step information seeking, processing, and reasoning to generate comprehensive reports. In contrast to existing…
Beyond Surface Artifacts: Capturing Shared Latent Forgery Knowledge Across Modalities
Jingtong Dou, Chuancheng Shi, Jian Wang +3
As generative artificial intelligence evolves, deepfake attacks have escalated from single-modality manipulations to complex, multimodal threats. Existing forensic techniques face…
Who Transfers Safety? Identifying and Targeting Cross-Lingual Shared Safety Neurons
Xianhui Zhang, Chengyu Xie, Linxia Zhu +6
Multilingual safety remains significantly imbalanced, leaving non-high-resource (NHR) languages vulnerable compared to robust high-resource (HR) ones. Moreover, the neural mechanis…
TraceRouter: Robust Safety for Large Foundation Models via Path-Level Intervention
Chuancheng Shi, Shangze Li, Wenjun Lu +5
Despite their capabilities, large foundation models (LFMs) remain susceptible to adversarial manipulation. Current defenses predominantly rely on the "locality hypothesis", suppres…
Reasoning-VLA: A Fast and General Vision-Language-Action Reasoning Model for Autonomous Driving
Dapeng Zhang, Zhenlong Yuan, Zhangquan Chen +5
Vision-Language-Action (VLA) models have recently shown strong decision-making capabilities in autonomous driving. However, existing VLAs often struggle with achieving efficient in…
Where Culture Fades: Revealing the Cultural Gap in Text-to-Image Generation
Chuancheng Shi, Shangze Li, Shiming Guo +9
Multilingual text-to-image (T2I) models have advanced rapidly in terms of visual realism and semantic alignment, and are now widely utilized. Yet outputs vary across cultural conte…