喜訊 | 48 篇入選！港科大（廣州）人工智能學域ICML 2026再創佳績，2篇Oral、2篇Spotlight

論文錄用結果

Acoustic Interference: A New Paradigm Weaponizing Acoustic Latent Semantic for Universal Jailbreak against Large Audio Language Models

作者：Yanyun WANG（AI MPhil 王彥云）, Yu HUANG（AI MPhil 黃宇）, Zi LIANG, Xixin WU, Li LIU（助理教授 劉李）

Agent-Omit: Adaptive Context Omission for Efficient LLM Agents

作者：Yansong NING（AI PhD 甯彥淞）, Jun FAN, Naiqiang TAN, Hao LIU（助理教授 劉浩）

AG-REPA: Causal Layer Selection for Representation Alignment in Audio Flow Matching

作者：Pengfei ZHANG（AI PhD 張鵬飛）, Tianxin XIE（AI PhD 謝添鑫）, Minghao YANG（AI PhD 楊銘皓）, Li LIU（助理教授 劉李）

All Circuits Lead to Rome: Rethinking Functional Anisotropy in Circuit and Sheaf Discovery for LLMs

作者：Xi CHEN, Mingyu JIN, Jingcheng NIU, Yutong YIN, Jinman ZHAO, Bangwei GUO, Dimitris N. METAXAS, Zhaoran WANG, Yutao YUE（副教授 岳玉濤）, Gerald PENN

AnyEdit++: Adaptive Long-Form Knowledge Editing via Bayesian Surprise

作者：Bowen TIAN（AI PhD 田博聞）, Caixue HE, Jiemin WU, Jingying WANG, Wenshuo CHEN, Zexi LI, Yutao YUE（副教授 岳玉濤）

ArcDAE: Asymmetric Rectified Contrastive Diffusion Autoencoder for Unified Representation Learning

作者：Ge GAO, Di XIONG, Zeke XIE（助理教授 謝澤柯）, Jian YANG, Shuo CHEN

DRIFT: Decoupled Rollouts and Importance-Weighted Fine-Tuning for Efficient Multi-Turn Optimization

作者：Jian MU（RA）, Tianyi LIN, Chengwei QIN（助理教授 秦成偉）, Zhongxiang DAI, Yao SHU（助理教授 舒瑤）

Exploring Data-Free LoRA Transferability for Video Diffusion Models

作者：Yuchen WANG, Wenliang ZHONG（AI PhD 鐘文亮）, Lichen BAI（AI PhD 白立忱）, Zikai ZHOU, Shitong SHAO, Bojun CHENG, Shuo CHEN, Shuo YANG, Zeke XIE（助理教授 謝澤柯）

Find, Fix, Reason: Context Repair for Video Reasoning

作者：Haojian HUANG（AI PhD 黃浩健）, Chuanyu QIN, Yinchuan LI, Yingcong CHEN（助理教授 陳穎聰）

FlatLand: Personalized Graph Federated Learning via Tailored Lorentz Space(Oral,&& Spotlight)

作者：Jiahong LIU, Ram Samarth B B, Xinyu FU, Menglin YANG（助理教授 楊萌林）, Weixi ZHANG, Rex YING, Irwin KING

Hierarchical Abstract Tree for Cross-Document Retrieval-Augmented Generation

作者：Ziwen ZHAO（AI PhD 趙子文）, Menglin YANG（助理教授 楊萌林）

Hyperbolic Multimodal Continual Learning

作者：Jiahong LIU, Ming SHEN, Xiaohao LIU, Rex YING, Menglin YANG（助理教授 楊萌林）, Tat-Seng CHUA, Irwin KING

HypRAG: Hyperbolic Dense Retrieval for Retrieval Augmented Generation

作者：Hiren MADHU, Ngoc BUI, Ali MAATOUK, Leandros TASSIULAS, Smita KRISHNASWAMY, Menglin YANG（助理教授 楊萌林）, Sukanta GANGULY, Kiran SRINIVASAN, Rex YING

Identifying and Mitigating Errors in Gradient Aggregation of Distributed Data Parallel Training

作者：Zhenheng TANG, Junlin HUANG, Zichen TANG,Xueze KANG, Yuxin WANG, Peijie DONG, Shaohuai SHI, Xiaowen CHU（教授 褚曉文）, Bo LI

ImpText: A Benchmark and Tool-Augmented Framework for Implicit Text Reasoning

作者：Litao GUO（AI PhD 郭力濤）, Jinsong ZHOU（AI MPhil 周勁松）, Shuaibo LI, Man CHEN（AI MPhil 陳曼）, Xinli XU（AI PhD 許新里）, Zixin ZHANG（AI PhD 張梓欣）, Harold Haodong CHEN（AI PhD 陳浩東）, Ying-Cong CHEN（助理教授 陳穎聰）

Information Flow Reveals When to Trust Language Models

作者：Rui XU（AI PhD，徐睿）, Yi CHEN, Jiujiu CHEN（AI PhD，陳玖玖）, Sihong XIE（副教授 謝思泓）

KineFlow: Kinematic Second-Order Flow Matching for Time-Series Forecasting

作者：Haiqi JIANG（AI MPhil 江海琪）, Hui XIONG（教授 熊輝）

Learning from Fine-Grained Visual Discrepancies: Mitigating Multimodal Hallucinations via In-Context Visual Contrastive Optimization

作者：Haolin DENG（AI PhD 鄧浩霖）, Xin ZOU（AI PhD 鄒鑫）, Zhiwei JIN, Chen CHEN, Haonan LU, Xuming HU（助理教授 胡旭明）

Learning Query-Aware Budget-Tier Routing for Runtime Agent Memory

作者：Haozhen ZHANG, Haodong YUE, Tao FENG, Quanyu LONG, Jianzhu BAO, Bowen JIN, Weizhi ZHANG, Xiao LI, Jiaxuan YOU, Chengwei QIN（助理教授 秦成偉）, Wenya WANG

Learning to Think in Physics: Breaking Shortcut Learning in Scientific Diffusion via Representation Alignment

作者：Haozhe JIA（RA）, Pengyu YIN, Wenshuo CHEN（AI PhD 陳文碩）, Shaofeng LIANG（RA）, Lei WANG（Postdoc）, Bowen TIAN（AI PhD 田博聞）, Xiucheng WANG, Nanqian JIA, Yutao YUE（副教授 岳玉濤）

Lightning Unified Video Editor via In-Context Sparse Attention

作者：Shitong SHAO, Zikai ZHOU, Haopeng LI（AI MPhil 李浩鵬）, Yingwei SONG, Wenliang ZHONG（AI PhD 鐘文亮）, Lichen BAI（AI PhD 白立忱）, Zeke XIE（助理教授 謝澤柯）

Low-cost Full Fine-tuning: Learning What to Update for LLMs

作者：Jin LI（AI PhD 李進）, Yaming GUO（AI PhD 郭亞明）, Shenghao GAO, Xinlong CHEN, Zuhao XU, Ying SUN, Chao WANG, Hui XIONG（教授 熊輝）

MASPOB: Bandit-Based Prompt Optimization for Multi-Agent Systems with Graph Neural Networks

作者：Zhi HONG, Qian ZHANG, Jiahang SUN, Zhiwei SHANG, Mingze KONG, Xiangyi WANG, Yao SHU（助理教授 舒瑤）, Zhongxiang DAI

MDN: Parallelizing Stepwise Momentum for Delta Linear Attention

作者：Yulong HUANG, Xiang LIU, Hongxiang HUANG, Xiaopeng LIN, Zunchang LIU, Xiaowen CHU（教授 褚曉文）, Zeke XIE（助理教授 謝澤柯）, Bojun CHENG

Med-Scout: Curing MLLMs' Geometric Blindness in Medical Perception via Geometry-Aware RL Post-Training

作者：Anglin LIU（AI PhD 劉昂霖）, Ruichao CHEN, Yi LU, Hongxia XU, Jintai CHEN（助理教授 陳晉泰）

MGAL: A Multilingual Granularity-Aware Long-Context Benchmark

作者：Chunhan LI, Chenglin XU, Zongyang ZHANG, Jiale LIU, Zhuoxi RAO, Xudong JIA, Junxiu HE, Menglin YANG（助理教授 楊萌林）, Wenjuan GONG, Zhengzhe LIU, Chengwei QIN（助理教授 秦成偉）

MotionGRPO: Overcoming Low Intra-Group Diversity in GRPO-Based Egocentric Motion Recovery

作者：Nanjie YAO（Incoming AI PhD）, Junlong REN（AI MPhil 任俊龍）, Wenhao SHEN, Hao WANG（助理教授 王浩）

On the Plasticity and Stability for Post-Training Large Language Models

作者：Wenwen QIANG, Ziyin GU, Jiahuan ZHOU, Jie HU, Jingyao WANG, Changwen ZHENG, Hui XIONG（教授 熊輝）

Op-CAD: Benchmarking and Investigating Operation-oriented CAD Generation

作者：Yixue BAI, Yufei GU, Zeke XIE（助理教授 謝澤柯）

Optimizing Few-Step Generation with Adaptive Matching Distillation

作者：Lichen BAI（AI PhD 白立忱）, Zikai ZHOU, Wenliang ZHONG（AI PhD 鐘文亮）, Shitong SHAO, Shuo YANG, Shuo CHEN, Bojun CHENG, Zeke XIE（助理教授 謝澤柯）

PSG-Nav: Probabilistic Scene Graph Navigation via Multiverse Decision Making

作者：Rufeng CHEN（AI PhD 陳如楓）, Yue CHANG（AI PhD 常越）, Xiaqiang TANG（AI PhD 唐夏強）, Hechang CHEN, Sihong XIE（副教授 謝思泓）

Revisiting Zeroth-Order Hessian Approximation: A Single-Step Policy Optimization Lens

作者：Junbin QIU（AI PhD 邱俊斌）, Zhaowei HONG（AI MPhil 洪兆煒）, Renzhe XU, Yao SHU（助理教授 舒瑤）

RLIE: Rule Generation with Logistic Regression, Iterative Refinement, and Evaluation for Large Language Models

作者：Yang YANG（AI PhD 楊洋）, Hua XU, Zhangyi HU（AI PhD 胡張藝）, Yutao YUE（副教授 岳玉濤）

SARSteer: Safeguarding Large Audio Language Models via Safe-Ablated Refusal Steering

作者：Weilin LIN（AI PhD 林威霖）, Jianze LI, Hui XIONG（教授 熊輝）, Li LIU（助理教授 劉李）

ScalingAR: Scaling Confidence for Autoregressive Image Generation

作者：Harold Haodong CHEN（AI PhD 陳浩東）, Xianfeng WU, Wen-Jie SHU, Rongjin GUO, Disen LAN, Harry YANG, Ying-Cong CHEN（助理教授 陳穎聰）

Semantic Integrity Matters: Benchmarking and Preserving High-Density Reasoning in KV Cache Compression

作者：Xiang Liu, Zhenheng Tang, Hong Chen（AI MPhil 陳鴻）, Peijie Dong, Zeyu Li, Xiuze Zhou（AI PhD 周秀澤）, Bo Li, Xuming Hu（助理教授 胡旭明）, Xiaowen Chu（教授 褚曉文）

Show, Don't Tell: Morphing Latent Reasoning into Image Generation

作者：Harold Haodong CHEN（AI PhD 陳浩東）, Xinxiang YIN, Wen-Jie SHU, Hongfei ZHANG, Zixin ZHANG（AI PhD 張梓欣）, Chenfei LIAO, Litao GUO（AI PhD 郭力濤）, Qifeng CHEN, Ying-Cong CHEN（助理教授 陳穎聰）

Spatial Memory for Out-of-Vision Manipulation in Vision-Language-Action

作者：Pengteng LI（AI PhD 李鵬騰）, Weiyu GUO（AI PhD 郭偉鈺）, He ZHANG, Tiefu CAI（AI PhD 蔡鐵夫）, Xiao HE, Yandong GUO, Hui XIONG（教授 熊輝）

StormInsight: Hierarchical Environmental Forcing and Vertical Coupling for Weather System Evolution

作者：Jun CHEN, Yan FANG, Minghui QIU, Yueran QIU, Lin CHEN, Shuxin ZHONG, Yu ZHANG,Kaishun WU（教授 伍楷舜）

The Forgetting-Retention Dilemma: Certified Unlearning Theory in Continual Learning

作者：Yiting HU, Lingjie DUAN（教授 段凌杰）, Qian ZHANG

The Perceptual Bandwidth Bottleneck in Vision-Language Models: Active Visual Reasoning via Sequential Experimental Design

作者：Anjie LIU（AI PhD 劉安杰）, Ziqin GONG（AI PhD 龔子欽）, Yan SONG, Yuxiang CHEN, Xiaolong LIU, Hengtong LU, Kaike ZHANG, Chen WEI, Jun WANG

Theory of Continual Learning Against Data Poisoning Attacks

作者：Yiting HU, Lingjie DUAN（教授 段凌杰）

Towards the Explainability of Temporal Graph Networks via Memory Backtracking and Topological Attribution

作者：Yazheng LIU（AI PhD 劉雅崢）, Xi ZHANG, Sihong XIE（副教授 謝思泓）, Hui XIONG（教授 熊輝）

T-POP: Test-Time Personalization with Online Preference Feedback

作者：Zikun QU, Min ZHANG, Mingze KONG, Xiang LI, Zhiwei SHANG, Zhiyong WANG, Yikun BAN, Shuang QIU, Yao SHU（助理教授 舒瑤）, Zhongxiang DAI

VCG-Bench: Towards A Unified Visual-Centric Benchmark for Structured Generation and Editing

作者：Xiaoyan SU, Peijie DONG, Zhenheng TANG, Song TANG, Yuyao ZHAI, Kaitao LIN, Liang CHEN, Yuhang GAI, Yuyu LUO, Qiang WANG, Xiaowen CHU（教授 褚曉文）

Video-MTR: Reinforced Multi-Turn Reasoning for Long Video Understanding

作者：Yuan XIE（AI PhD 謝圓）, Tianshui CHEN, Zheng GE, Lionel NI（教授 倪明選）

Words & Weights: Streamlining Multi-Turn Interactions via Co-Adaptation

作者：Chenxing WEI, Hong WANG, Ying HE, Zhongxiang DAI, Bo JIANG, F Richard YU, Yao SHU（助理教授 舒瑤）

X-EviProbe: Post-hoc Parameter-free Evidential Uncertainty Quantification for Frozen Graph Neural Networks

作者：Chenghua GUO,Sihong XIE（副教授 謝思泓）, Xi ZHANG