15 papers
Linear Fitness Subspace in Protein Language Models Enables Sample-Efficient Directed Evolution
SiYuan Ma, Canran Xiao, Zikai Xiao +5
Model-guided directed evolution seeks to identify high-fitness protein variants under limited oracle budgets. Protein language models (PLMs) provide rich representations for this t…
MCD: Causal Distillation of Multimodal In-Context Learning in Large Vision-Language Models
Yanshu Li, Jiaqian Li, Canran Xiao +3
Large vision-language models (LVLMs) exhibit strong multimodal in-context learning (ICL) capabilities, yet this ability degrades substantially as model size decreases. Knowledge di…
Who Bridges Safety? Identifying and Targeting Cross-Lingual Shared Safety Pathways
Shuyi Miao, Wangjie Qiu, Pengyang Shao +4
Uncovering the internal mechanisms underlying the safety capabilities of large language models (LLMs) is crucial for developing trustworthy artificial intelligence. Currently, mech…
Hidden APIs in Language Models: Discovering Reusable Causal Interfaces from Forked Futures
SiYuan Ma, Yiqin Luo, Zhangji +8
Identical language-model answers can arise from hidden states that support different future computations, so current-answer probes do not establish a reusable internal interface. W…
LaP-Forensics: Latent-Pixel Consistency Guided Multimodal Reasoning for Deepfake Detection
Can Wang, Yuhao Wang, Yushe Cao +2
Recent generative models can produce images with few obvious visual artifacts, weakening detectors and explanations that rely only on surface appearance. We present LaP-Forensics,…
Text as Partial Constraint: Core-Residual Alignment for Robust Vision-Language Learning
Chengzhen Yu, Canran Xiao, Siyuan Ma +1
Vision-language alignment powers open-vocabulary recognition, retrieval, and LVLM grounding, yet natural captions are often underspecified, making similarity brittle and overly con…
Hidden Forgetting in Continual Multimodal Learning: When Accuracy Survives but Grounding Fails
Qianyu Chen, Canran Xiao, Runxuan Tang
Multimodal large language models must continually adapt to evolving tasks and domains, yet standard continual learning metrics mainly measure whether old answers remain correct, le…
InduceKV: Fixed-Footprint Continual Adaptation of Multimodal LLMs via Inducing KV Memories
Qianyu Chen, Ziteng Feng, Canran Xiao +1
Multimodal large language models must adapt to evolving tasks and domains, yet continual improvement under bounded deployment footprint remains difficult because repeated parameter…
HG-Bench: A Benchmark for Multi-Page Handwritten Answer-Region Grounding in Automated Homework Assessment
Chuangxin Zhao, Boyan Shi, Yanling Wang +7
Automated homework assessment depends not only on recognizing student answers, but also on accurately locating where each answer and each intermediate reasoning step appears in noi…
Attention-Spectrum Regularization for Replay-Free Continual Multimodal LLMs
Chuangxin Zhao, Canran Xiao, Siyuan Ma +5
Multimodal large language models (MLLMs) are increasingly required to adapt to non-stationary streams of visual domains, question types, and user instructions, yet continual fine-t…
How to Achieve Prototypical Birth and Death for OOD Detection?
Ningkang Peng, Qianfeng Yu, Xiaoqian Peng +7
Out-of-Distribution (OOD) detection is crucial for the secure deployment of machine learning models, and prototype-based learning methods are among the mainstream strategies for ac…
Seeing through the Conflict: Transparent Knowledge Conflict Handling in Retrieval-Augmented Generation
Hua Ye, Siyuan Chen, Ziqi Zhong +4
Large language models (LLMs) equipped with retrieval--the Retrieval-Augmented Generation (RAG) paradigm--should combine their parametric knowledge with external evidence, yet in pr…
Affordance-First Decomposition for Continual Learning in Video-Language Understanding
Mengzhu Xu, Hanzhi Liu, Ningkang Peng +2
Continual learning for video--language understanding is increasingly important as models face non-stationary data, domains, and query styles, yet prevailing solutions blur what sho…
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…
C3-OWD: A Curriculum Cross-modal Contrastive Learning Framework for Open-World Detection
Siheng Wang, Zhengdao Li, Yanshu Li +12
Object detection has advanced significantly in the closed-set setting, but real-world deployment remains limited by two challenges: poor generalization to unseen categories and ins…