Publications

2026

1. ECCV

   LISA: Locality-Informed Speculative Decoding for Accelerating Autoregressive Image Generation

   Ying Li, Siyong Jian, Zhaode Wang, Zhiwen Chen, Chengfei Lv, and Huan Wang

   In European Conference on Computer Vision (ECCV), 2026

   Bib

   @inproceedings{li2026lisa,
     title = {LISA: Locality-Informed Speculative Decoding for Accelerating Autoregressive Image Generation},
     author = {Li, Ying and Jian, Siyong and Wang, Zhaode and Chen, Zhiwen and Lv, Chengfei and Wang, Huan},
     booktitle = {European Conference on Computer Vision (ECCV)},
     year = {2026},
   }

2. ECCV

   EVAR: Edge Visual Autoregressive Models via Principled Pruning

   Zefang Wang, Ying Li, Yanyu Li, Mingluo Su, Simin Xu, Guanzhong Tian, and Huan Wang

   In European Conference on Computer Vision (ECCV), 2026

   Bib

   @inproceedings{wang2026evar,
     title = {EVAR: Edge Visual Autoregressive Models via Principled Pruning},
     author = {Wang, Zefang and Li, Ying and Li, Yanyu and Su, Mingluo and Xu, Simin and Tian, Guanzhong and Wang, Huan},
     booktitle = {European Conference on Computer Vision (ECCV)},
     year = {2026},
   }

3. ECCV

   TARS: MinMax Token-Adaptive Preference Strategy for Hallucination Reduction in MLLMs

   Kejia Zhang, Keda Tao, Zhiming Luo, Chang Liu, Jiasheng Tang, and Huan Wang

   In European Conference on Computer Vision (ECCV), 2026

   arXiv Bib PDF Code Website

   @inproceedings{zhang2026tars,
     title = {TARS: MinMax Token-Adaptive Preference Strategy for Hallucination Reduction in MLLMs},
     author = {Zhang, Kejia and Tao, Keda and Luo, Zhiming and Liu, Chang and Tang, Jiasheng and Wang, Huan},
     booktitle = {European Conference on Computer Vision (ECCV)},
     year = {2026},
   }

4. ICML

   ARC-Decode: Accelerated Decoding with Risk-Bounded Acceptance

   Ying Li, Zhaode Wang, Zhiwen Chen, Chengfei Lv, and Huan Wang

   In International Conference on Machine Learning (ICML), 2026

   Bib

   @inproceedings{li2026arcdecode,
     title = {ARC-Decode: Accelerated Decoding with Risk-Bounded Acceptance},
     author = {Li, Ying and Wang, Zhaode and Chen, Zhiwen and Lv, Chengfei and Wang, Huan},
     booktitle = {International Conference on Machine Learning (ICML)},
     year = {2026},
   }

5. ICML

   Prism-MoE: Efficient Dense-to-MoE Conversion for Visual Autoregressive Generation

   Ying Li, Zefang Wang, Zhaode Wang, Zhiwen Chen, Chengfei Lv, and Huan Wang

   In International Conference on Machine Learning (ICML), 2026

   Bib

   @inproceedings{li2026prismmoe,
     title = {Prism-MoE: Efficient Dense-to-MoE Conversion for Visual Autoregressive Generation},
     author = {Li, Ying and Wang, Zefang and Wang, Zhaode and Chen, Zhiwen and Lv, Chengfei and Wang, Huan},
     booktitle = {International Conference on Machine Learning (ICML)},
     year = {2026},
   }

6. ICML

   Which Heads Matter for Reasoning? RL-Guided KV Cache Compression

   Wenjie Du, Li Jiang, Keda Tao, Xue Liu, and Huan Wang

   In International Conference on Machine Learning (ICML), 2026

   arXiv Bib PDF Code Website

   @inproceedings{du2026rlkv,
     title = {Which Heads Matter for Reasoning? RL-Guided KV Cache Compression},
     author = {Du, Wenjie and Jiang, Li and Tao, Keda and Liu, Xue and Wang, Huan},
     booktitle = {International Conference on Machine Learning (ICML)},
     year = {2026},
   }

7. ICML

   SparseSSM: Efficient Selective Structured State Space Models Can Be Pruned in One-Shot

   Kaiwen Tuo and Huan Wang

   In International Conference on Machine Learning (ICML), 2026

   arXiv Bib PDF Code Website

   @inproceedings{tuo2026sparsessm,
     title = {SparseSSM: Efficient Selective Structured State Space Models Can Be Pruned in One-Shot},
     author = {Tuo, Kaiwen and Wang, Huan},
     booktitle = {International Conference on Machine Learning (ICML)},
     year = {2026},
   }

8. CVPR

   Parallel Jacobi Decoding for Fast Autoregressive Image Generation

   Boya Liao, Ying Li, Siyong Jian, and Huan Wang

   In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026

   arXiv Bib PDF

   @inproceedings{liao2026jacobi,
     title = {Parallel Jacobi Decoding for Fast Autoregressive Image Generation},
     author = {Liao, Boya and Li, Ying and Jian, Siyong and Wang, Huan},
     booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
     year = {2026},
   }

9. CVPR

   QVGGT: Post-Training Quantized Visual Geometry Grounded Transformer

   Zhizhen Pan, Hesong Wang, and Huan Wang

   In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026

   arXiv Bib PDF

   @inproceedings{pan2026qvggt,
     title = {QVGGT: Post-Training Quantized Visual Geometry Grounded Transformer},
     author = {Pan, Zhizhen and Wang, Hesong and Wang, Huan},
     booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
     year = {2026},
   }

10. CVPR

    EarlyTom: Early Token Compression Completes Fast Video Understanding

    Hesong Wang, Xin Jin, Lu Lu, Chenhaowen Li, Jian Chen, Qiang Liu, and Huan Wang

    In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026

    arXiv Bib PDF

    @inproceedings{wang2026earlytom,
      title = {EarlyTom: Early Token Compression Completes Fast Video Understanding},
      author = {Wang, Hesong and Jin, Xin and Lu, Lu and Li, Chenhaowen and Chen, Jian and Liu, Qiang and Wang, Huan},
      booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
      year = {2026},
    }

11. CVPR

    OmniZip: Audio-Guided Dynamic Token Compression for Fast Omnimodal Large Language Models

    Keda Tao, Kele Shao, Bohan Yu, Weiqiang Wang, Jian Liu, and Huan Wang

    In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026

    arXiv Bib PDF Code

    @inproceedings{tao2026omnizip,
      title = {OmniZip: Audio-Guided Dynamic Token Compression for Fast Omnimodal Large Language Models},
      author = {Tao, Keda and Shao, Kele and Yu, Bohan and Wang, Weiqiang and Liu, Jian and Wang, Huan},
      booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
      year = {2026},
    }

12. CVPR

    StreamingTOM: Streaming Token Compression for Efficient Video Understanding

    Xueyi Chen, Keda Tao, Kele Shao, and Huan Wang

    In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026

    arXiv Bib PDF Code Website

    @inproceedings{chen2026streamingtom,
      title = {StreamingTOM: Streaming Token Compression for Efficient Video Understanding},
      author = {Chen, Xueyi and Tao, Keda and Shao, Kele and Wang, Huan},
      booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
      year = {2026},
    }

13. CVPR

    Can MLLMs Guide Me Home? A Benchmark Study on Fine-Grained Visual Reasoning from Transit Maps

    Sicheng Feng, Song Wang, Shuyi Ouyang, Lingdong Kong, Zikai Song, Jianke Zhu, Huan Wang, and Xinchao Wang

    In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026

    arXiv Bib PDF Code Website

    @inproceedings{feng2026reasonmap,
      title = {Can MLLMs Guide Me Home? A Benchmark Study on Fine-Grained Visual Reasoning from Transit Maps},
      author = {Feng, Sicheng and Wang, Song and Ouyang, Shuyi and Kong, Lingdong and Song, Zikai and Zhu, Jianke and Wang, Huan and Wang, Xinchao},
      booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
      year = {2026},
    }

14. ICLR

    MergeMix: A Unified Augmentation Paradigm for Visual and Multi-Modal Understanding

    Xin Jin, Siyuan Li, Siyong Jian, Kai Yu, and Huan Wang

    In International Conference on Learning Representations (ICLR), 2026

    arXiv Bib PDF Code Website

    @inproceedings{jin2026mergemix,
      title = {MergeMix: A Unified Augmentation Paradigm for Visual and Multi-Modal Understanding},
      author = {Jin, Xin and Li, Siyuan and Jian, Siyong and Yu, Kai and Wang, Huan},
      booktitle = {International Conference on Learning Representations (ICLR)},
      year = {2026},
    }

15. ICLR

    OBS-Diff: Accurate Pruning For Diffusion Models in One-Shot

    Junhan Zhu, Hesong Wang, Mingluo Su, Zefang Wang, and Huan Wang

    In International Conference on Learning Representations (ICLR), 2026

    arXiv Bib PDF Code Website

    @inproceedings{zhu2026obsdiff,
      title = {OBS-Diff: Accurate Pruning For Diffusion Models in One-Shot},
      author = {Zhu, Junhan and Wang, Hesong and Su, Mingluo and Wang, Zefang and Wang, Huan},
      booktitle = {International Conference on Learning Representations (ICLR)},
      year = {2026},
    }

16. ICLR

    RewardMap: Tackling Sparse Rewards in Fine-grained Visual Reasoning via Multi-Stage Reinforcement Learning

    Sicheng Feng, Kaiwen Tuo, Song Wang, Lingdong Kong, Jianke Zhu, and Huan Wang

    In International Conference on Learning Representations (ICLR), 2026

    arXiv Bib PDF Code Website

    @inproceedings{feng2026rewardmap,
      title = {RewardMap: Tackling Sparse Rewards in Fine-grained Visual Reasoning via Multi-Stage Reinforcement Learning},
      author = {Feng, Sicheng and Tuo, Kaiwen and Wang, Song and Kong, Lingdong and Zhu, Jianke and Wang, Huan},
      booktitle = {International Conference on Learning Representations (ICLR)},
      year = {2026},
    }

17. ICLR

    Autoregressive Image Generation with Randomized Parallel Decoding

    Haopeng Li, Jinyue Yang, Guoqi Li, and Huan Wang

    In International Conference on Learning Representations (ICLR), 2026

    arXiv Bib PDF Code Website

    @inproceedings{li2026arpg,
      title = {Autoregressive Image Generation with Randomized Parallel Decoding},
      author = {Li, Haopeng and Yang, Jinyue and Li, Guoqi and Wang, Huan},
      booktitle = {International Conference on Learning Representations (ICLR)},
      year = {2026},
    }

18. CPAL Oral

    ROSE: Reordered SparseGPT for More Accurate One-Shot Large Language Models Pruning

    Mingluo Su and Huan Wang

    In Conference on Parsimony and Learning (CPAL), 2026

    arXiv Bib PDF Code Website

    @inproceedings{su2026rose,
      title = {ROSE: Reordered SparseGPT for More Accurate One-Shot Large Language Models Pruning},
      author = {Su, Mingluo and Wang, Huan},
      booktitle = {Conference on Parsimony and Learning (CPAL)},
      year = {2026},
    }

19. CPAL

    ResSVD: Residual Compensated SVD for Large Language Model Compression

    Haolei Bai, Siyong Jian, Tuo Liang, Yu Yin, and Huan Wang

    In Conference on Parsimony and Learning (CPAL), 2026

    arXiv Bib PDF Code

    @inproceedings{bai2026ressvd,
      title = {ResSVD: Residual Compensated SVD for Large Language Model Compression},
      author = {Bai, Haolei and Jian, Siyong and Liang, Tuo and Yin, Yu and Wang, Huan},
      booktitle = {Conference on Parsimony and Learning (CPAL)},
      year = {2026},
    }

20. TMLR

    When Tokens Talk Too Much: A Survey of Multimodal Long-Context Token Compression across Images, Videos, and Audios

    Kele Shao, Keda Tao, Kejia Zhang, Sicheng Feng, Mu Cai, Yuzhang Shang, Haoxuan You, Can Qin, Yang Sui, and Huan Wang

    Transactions on Machine Learning Research (TMLR), 2026

    arXiv Bib PDF Code

    @article{shao2026tokens,
      title = {When Tokens Talk Too Much: A Survey of Multimodal Long-Context Token Compression across Images, Videos, and Audios},
      author = {Shao, Kele and Tao, Keda and Zhang, Kejia and Feng, Sicheng and Cai, Mu and Shang, Yuzhang and You, Haoxuan and Qin, Can and Sui, Yang and Wang, Huan},
      journal = {Transactions on Machine Learning Research (TMLR)},
      year = {2026},
    }

21. TMLR

    Is Oracle Pruning the True Oracle?

    Sicheng Feng, Keda Tao, and Huan Wang

    Transactions on Machine Learning Research (TMLR), 2026

    arXiv Bib PDF Code Website

    @article{feng2026oracle,
      title = {Is Oracle Pruning the True Oracle?},
      author = {Feng, Sicheng and Tao, Keda and Wang, Huan},
      journal = {Transactions on Machine Learning Research (TMLR)},
      year = {2026},
    }

22. arXiv

    LVOmniBench: Pioneering Long Audio-Video Understanding Evaluation for Omnimodal LLMs

    Keda Tao, Yuhua Zheng, Jia Xu, Wenjie Du, Kele Shao, Hesong Wang, Xueyi Chen, Xin Jin, Junhan Zhu, Bohan Yu, Weiqiang Wang, Jian Liu, Can Qin, Yulun Zhang, Ming-Hsuan Yang, and Huan Wang

    arXiv preprint arXiv:2603.19217, 2026

    arXiv Bib PDF Code Website

    @article{tao2026lvomnibench,
      title = {LVOmniBench: Pioneering Long Audio-Video Understanding Evaluation for Omnimodal LLMs},
      author = {Tao, Keda and Zheng, Yuhua and Xu, Jia and Du, Wenjie and Shao, Kele and Wang, Hesong and Chen, Xueyi and Jin, Xin and Zhu, Junhan and Yu, Bohan and Wang, Weiqiang and Liu, Jian and Qin, Can and Zhang, Yulun and Yang, Ming-Hsuan and Wang, Huan},
      journal = {arXiv preprint arXiv:2603.19217},
      year = {2026},
    }

23. arXiv

    MobileKernelBench: Can LLMs Write Efficient Kernels for Mobile Devices?

    Xingze Zou, Jing Wang, Yuhua Zheng, Xueyi Chen, Haolei Bai, Lingcheng Kong, Syed A.R. Abu-Bakar, Zhaode Wang, Chengfei Lv, Haoji Hu, and Huan Wang

    arXiv preprint arXiv:2603.11935, 2026

    arXiv Bib PDF Code Website

    @article{zou2026mobilekernelbench,
      title = {MobileKernelBench: Can LLMs Write Efficient Kernels for Mobile Devices?},
      author = {Zou, Xingze and Wang, Jing and Zheng, Yuhua and Chen, Xueyi and Bai, Haolei and Kong, Lingcheng and Abu-Bakar, Syed A.R. and Wang, Zhaode and Lv, Chengfei and Hu, Haoji and Wang, Huan},
      journal = {arXiv preprint arXiv:2603.11935},
      year = {2026},
    }

24. arXiv

    DICE: Diffusion Large Language Models Excel at Generating CUDA Kernels

    Haolei Bai, Lingcheng Kong, Xueyi Chen, Jianmian Wang, Zhiqiang Tao, and Huan Wang

    arXiv preprint arXiv:2602.11715, 2026

    arXiv Bib PDF Code Website

    @article{bai2026dice,
      title = {DICE: Diffusion Large Language Models Excel at Generating CUDA Kernels},
      author = {Bai, Haolei and Kong, Lingcheng and Chen, Xueyi and Wang, Jianmian and Tao, Zhiqiang and Wang, Huan},
      journal = {arXiv preprint arXiv:2602.11715},
      year = {2026},
    }

2025

1. NeurIPS

   FreqExit: Enabling Early-Exit Inference for Visual Autoregressive Models via Frequency-Aware Guidance

   Ying Li, Chengfei Lv, and Huan Wang

   In Advances in Neural Information Processing Systems (NeurIPS), 2025

   Bib PDF Code Website

   @inproceedings{li2025freqexit,
     title = {FreqExit: Enabling Early-Exit Inference for Visual Autoregressive Models via Frequency-Aware Guidance},
     author = {Li, Ying and Lv, Chengfei and Wang, Huan},
     booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
     year = {2025},
   }

2. NeurIPS

   HoliTom: Holistic Token Merging for Fast Video Large Language Models

   Kele Shao, Keda Tao, Can Qin, Haoxuan You, Yang Sui, and Huan Wang

   In Advances in Neural Information Processing Systems (NeurIPS), 2025

   arXiv Bib PDF Code Website

   @inproceedings{shao2025holitom,
     title = {HoliTom: Holistic Token Merging for Fast Video Large Language Models},
     author = {Shao, Kele and Tao, Keda and Qin, Can and You, Haoxuan and Sui, Yang and Wang, Huan},
     booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
     year = {2025},
   }

3. NeurIPS

   Poison as Cure: Visual Noise for Mitigating Object Hallucinations in LVMs

   Kejia Zhang, Keda Tao, Jiasheng Tang, and Huan Wang

   In Advances in Neural Information Processing Systems (NeurIPS), 2025

   arXiv Bib PDF Code Website

   @inproceedings{zhang2025poisoncure,
     title = {Poison as Cure: Visual Noise for Mitigating Object Hallucinations in LVMs},
     author = {Zhang, Kejia and Tao, Keda and Tang, Jiasheng and Wang, Huan},
     booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
     year = {2025},
   }

4. CVPR

   DyCoke: Dynamic Compression of Tokens for Fast Video Large Language Models

   Keda Tao, Can Qin, Haoxuan You, Yang Sui, and Huan Wang

   In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025

   arXiv Bib PDF Code

   @inproceedings{tao2025dycoke,
     title = {DyCoke: Dynamic Compression of Tokens for Fast Video Large Language Models},
     author = {Tao, Keda and Qin, Can and You, Haoxuan and Sui, Yang and Wang, Huan},
     booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
     year = {2025},
   }

5. TCSVT

   Niagara: Normal-Integrated Geometric Affine Field for Scene Reconstruction from a Single View

   Xianzu Wu, Zhenxin Ai, Harry Yang, Ser-Nam Lim, Jun Liu, and Huan Wang

   IEEE Transactions on Circuits and Systems for Video Technology, 2025

   arXiv Bib PDF Code Website

   @article{wu2025niagara,
     title = {Niagara: Normal-Integrated Geometric Affine Field for Scene Reconstruction from a Single View},
     author = {Wu, Xianzu and Ai, Zhenxin and Yang, Harry and Lim, Ser-Nam and Liu, Jun and Wang, Huan},
     journal = {IEEE Transactions on Circuits and Systems for Video Technology},
     year = {2025},
   }

6. arXiv

   Active Perception Agent for Omnimodal Audio-Video Understanding

   Keda Tao, Wenjie Du, Bohan Yu, Weiqiang Wang, Jian Liu, and Huan Wang

   arXiv preprint arXiv:2512.23646, 2025

   arXiv Bib PDF Code Website

   @article{tao2025omniagent,
     title = {Active Perception Agent for Omnimodal Audio-Video Understanding},
     author = {Tao, Keda and Du, Wenjie and Yu, Bohan and Wang, Weiqiang and Liu, Jian and Wang, Huan},
     journal = {arXiv preprint arXiv:2512.23646},
     year = {2025},
   }

7. arXiv

   ConCuR: Conciseness Makes State-of-the-Art Kernel Generation

   Lingcheng Kong, Jiateng Wei, Hanzhang Shen, and Huan Wang

   arXiv preprint arXiv:2510.07356, 2025

   arXiv Bib PDF

   @article{kong2025concur,
     title = {ConCuR: Conciseness Makes State-of-the-Art Kernel Generation},
     author = {Kong, Lingcheng and Wei, Jiateng and Shen, Hanzhang and Wang, Huan},
     journal = {arXiv preprint arXiv:2510.07356},
     year = {2025},
   }

8. arXiv

   Plug-and-Play 1.x-Bit KV Cache Quantization for Video Large Language Models

   Keda Tao, Haoxuan You, Yang Sui, Can Qin, and Huan Wang

   arXiv preprint arXiv:2503.16257, 2025

   arXiv Bib PDF Code

   @article{tao2025plugandplay,
     title = {Plug-and-Play 1.x-Bit KV Cache Quantization for Video Large Language Models},
     author = {Tao, Keda and You, Haoxuan and Sui, Yang and Qin, Can and Wang, Huan},
     journal = {arXiv preprint arXiv:2503.16257},
     year = {2025},
   }