王培松，中国科学院自动化研究所，副研究员，硕导


主要研究方向包括多模态大模型架构与计算优化、神经网络加速与压缩等，在IEEE TPAMI、NeurIPS、ICML、CVPR、ICCV、ECCV等国际期刊和会议发表论文50余篇。主持国家自然科学基金、联合基金课题、中科院战略先导子课题、科技创新2030子课题，以及华为、百度、OPPO等多项科研合作项目。曾获Nvidia奖学金、IEEE国际标准突出贡献奖、NeurIPS国际神经网络压缩竞赛MicroNet Challenge冠军，入选中科院特聘研究骨干、北京市科协青年人才托举工程、微软亚洲研究院“铸星计划”、CCF-百度松果基金学者等。

  • 2025 华为计算产品线优秀合作项目奖
  • 2024 电子工业出版社2024年度优秀作者奖
  • 2023 入选CCF-百度松果基金学者
  • 2023 入选中科院特聘研究骨干
  • 2022 入选北京科协青年人才托举工程
  • 2022 IEEE国际标准突出贡献奖
  • 2021 入选微软亚洲研究院“铸星计划”
  • 2019 NeurIPS 2019国际神经网络压缩挑战赛MicroNet Challenge冠军
  • 2018 NVIDIA奖学金

电子邮件： peisong.wang@nlpr.ia.ac.cn

团队主页：https://clab.ia.ac.cn/

代码和数据库：https://github.com/CAS-CLab

通信地址： 北京市海淀区中关村东路95号

招生信息

招生专业：

081104-模式识别与智能系统

081203-计算机应用技术


招生方向：

多模态大模型、计算机视觉、大语言模型、模型加速与压缩

团队长期招聘优秀实习生，如果有准备保研到课题组或准备出国的大二、大三学生，欢迎联系！

教育背景

2013-09--2018-06   中国科学院自动化研究所   工学博士
2009-09--2013-06   山东大学   工学学士

工作经历

2021-04~现在, 中国科学院自动化研究所 副研究员
2021-11~2022-05,微软亚洲研究院, “铸星计划”访问学者
2018-07~2021-04,中国科学院自动化研究所 助理研究员

出版信息


发表论文
  1. Yuantian Shao, Peisong Wang, Yuanteng Chen, Chang Xu, Zhihui Wei, Jian Cheng. Block Rotation is All You Need for MXFP4 Quantization. ICML 2026.
  2. Yuanteng Chen, Peisong Wang, Nanxin Zeng, Yuantian Shao, Shuang Qiu, Gang Li, Jing Liu, Jian Cheng. Certain Head, Uncertain Tail: Expert-Sample for Test-Time Scaling in Fine-Grained MoE. ICML 2026.
  3. Zhongjian Qiao, Jiafei Lyu, Chenjia Bai, Peisong Wang, Siyang Gao, Shuang Qiu. Unifying Value Alignment and Assignment in Cross-Domain Offline Reinforcement Learning with Heterogeneous Datasets. ICML 2026.
  4. Zekun Li, Ning Wang, Tongxin Bai, Changwang Mei, Peisong Wang, Shuang Qiu, Jian Cheng. SparVAR: Exploring Sparsity in Visual AutoRegressive Modeling for Training-Free Acceleration. CVPR 2026.
  5. Zhinan Xie, Peisong Wang, Shuang Qiu, Jian Cheng. HiViS: Hiding Visual Tokens from the Drafter for Speculative Decoding in Vision-Language Models. Findings of CVPR 2026.
  6. Renxing Chen, Ziwei Xiang, Peisong Wang, Hongjian Fang, Meng Li, Fanhu Zeng, Yanan Zhu, Peipei Yang, Xu-Yao Zhang, Jian Cheng. FARSS: Fisher-Optimized Adaptive Low-Rank and Singular-Vector Selection for Knowledge-Preserving Fine-Tuning. Findings of ACL 2026.
  7. Zeyu Zhu, Gang Li, Minnan Pei, Zitao Mo, Peisong Wang, Tielong Liu, Jian Cheng. KL-MoE: A Hierarchical MoE Pruning Framework Exploiting KL Divergence. DAC 2026.
  8. Yuanhui Wang, Kunlong Liu, Minnan Pei, Zhangming Li, Peisong Wang, Qinghao Hu. MemeBQ: Memory Efficient Binary Quantization of LLMs. AAAI 2026.
  9. Piehuan Ni, Zitao Mo, Tielong Liu, Hongli Wen, Zeyu Zhu, Minnan Pei, Junwen Si, Weifan Guan, Peisong Wang, Qinghao Hu, Gang Li, Jian Cheng. APEX: Integer-only Non-linear Function Approximation for Efficient Cross-Modal Inference. DATE 2026.
  10. Changwang Mei, Peisong Wang, Shuang Qiu, Gang Li, Qinghao Hu, Yifan Zhang, Zhihui Wei, Jian Cheng. HiCache: Hierarchical Timestep-Aware Caching for Diffusion Transformer Acceleration. Visual Intelligence, 2026.
  11. Tianqi Chen, Zhe Li, Peisong Wang, Weixiang Xu, Zeyu Zhu, Jian Cheng. TernaryLLM: Ternarized Large Language Model. Machine Intelligence Research (MIR), 2026.
  12. Jiahe Qian, Peisong Wang, Zhengyang Zhuge, Qinghao Hu, Jian Cheng. A Universal Self-Attention Enhancement for Bridging Low-bit Quantization and Vision Transformers. WACV 2026.
  13. Yuantian Shao, Yuanteng Chen, Peisong Wang, Jianlin Yu, Jing Lin, Yiwu Yao, Zhihui Wei, Jian Cheng. DartQuant: Efficient Rotational Distribution Calibration for LLM Quantization. NeurIPS 2025.
  14. Yuanteng Chen, Yuantian Shao, Peisong Wang, Jian Cheng. EAC-MoE: Expert-Selection Aware Compressor for Mixture-of-Experts Large Language Models. ACL 2025.
  15. Chen Tianqi, Yuanteng Chen, Peisong Wang, Weixiang Xu, Zeyu Zhu, Jian Cheng. Q-Mamba: Towards more efficient Mamba models via post-training quantization. Findings of ACL 2025.
  16. Chen Tianqi, Peisong Wang, Weixiang Xu, Zeyu Zhu, Jian Cheng. RQT: Hierarchical Residual Quantization for Multi-Model Compression. Findings of ACL 2025.
  17. Yingying Deng, Xiangyu He, Changwang Mei, Peisong Wang, Fan Tang. FireFlow: Fast Inversion of Rectified Flow for Image Semantic Editing. ICML 2025.
  18. Minnan Pei, Gang Li, Junwen Si, Zeyu Zhu, Zitao Mo, Peisong Wang, Zhuoran Song, Xiaoyao Liang, Jian Cheng. GCC: A 3DGS Inference Architecture with Gaussian-Wise and Cross-Stage Conditional Processing. MICRO 2025.
  19. Zhangming Li, Qinghao Hu, Yiqun Chen, Peisong Wang, Yifan Zhang, Jian Cheng. LoRaDA: Low-Rank Direct Attention Adaptation for Efficient LLM Fine-tuning. EMNLP 2025.
  20. Xingting Yao, Qinghao Hu, Fei Zhou, Tielong Liu, Gang Li, Peisong Wang, Jian Cheng. Towards Efficient and Accurate Spiking Neural Networks via Adaptive Bit Allocation. Neural Networks, 2025.
  21. Shengxiao Zhou, Chenghua Li, Jianhao Huang, Peisong Wang. Perspective-Aware Inpainting: Bridging Large Viewpoint Gaps for Reference-Guided Inpainting. International Conference on Virtual Reality and Visualization (ICVRV), 2025.
  22. Zhengyang Zhuge, JiaxingWang, Yong Li, Yongjun Bao, PeisongWang, Jian Cheng, Patch-Aware Sample Selection for Efficient Masked Image Modeling, AAAI 2024.
  23. Zhengyang Zhuge, Peisong Wang, Xingting Yao, Jian Cheng. Towards Efficient Spiking Transformer: a Token Sparsification Framework for Training and Inference Acceleration. ICML 2024. 
  24. Zeyu Zhu, Peisong Wang, Qinghao Hu, Gang Li, Xiaoyao Liang, Jian Cheng. FastGL: A GPU-Efficient Framework for Accelerating Sampling-Based GNN Training at Large Scale. ASPLOS 2024.
  25. Peisong Wang, Weihan Chen, Xiangyu He, Qiang Chen, Qingshan Liu, Jian Cheng. Optimization-based Post-training Quantization with Bit-split and Stitching. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2023.
  26. Weihan Chen, Peisong Wang, Jian Cheng. Towards automatic model compression via a unified two-stage framework. Pattern Recognation (PR), 2023. 
  27. Tianqi Chen, Weixiang Xu, Weihan Chen, Peisong Wang, Jian Cheng. Towards Efficient and Accurate Winograd Convolution via Full Quantization. NeurIPS 2023. 
  28. Xiaoran Qin, Yu Zhu, Chenghua Li, Peisong Wang, Jian Cheng. CIDBNet: a consecutively-interactive dual-branch network for JPEG compressed image super-resolution. ECCV 2022 workshop.
  29. Xiaoyi Dong, Yu Zhu, Chenghua Li, Peisong Wang, Jian Cheng. RISPNet: A network for Reversed image signal processing. ECCV 2022 workshop.
  30. Marcos V Conde, Radu Timofte, Yibin Huang, Jingyang Peng, Chang Chen, Cheng Li, Eduardo Pérez-Pellitero, Fenglong Song, Furui Bai, Shuai Liu, Chaoyu Feng, Xiaotao Wang, Lei Lei, Yu Zhu, Chenghua Li, Yingying Jiang, Yong A, Peisong Wang, Cong Leng, Jian Cheng, Xiaoyu Liu, Zhicun Yin, Zhilu Zhang, Junyi Li, Ming Liu, Wangmeng Zuo, Jun Jiang, Jinha Kim, Yue Zhang, Beiji Zou, Zhikai Zong, Xiaoxiao Liu, Juan Marín Vega, Michael Sloth, Peter Schneider-Kamp, Richard Röttger, Furkan Kınlı, Barış Özcan, Furkan Kıraç, Li Leyi, SM Nadim Uddin, Dipon Kumar Ghosh, Yong Ju Jung. Reversed image signal processing and RAW reconstruction. AIM 2022 challenge report. ECCV 2022 workshop.
  31. Peisong Wang*, Xiangyu He*, Jian Cheng. Towards Accurate Binarized Neural Networks with Sparsity for Mobile Application. IEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2022.
  32. Weixiang Xu*, Fanrong Li*, Yingying Jiang, Yong A, Xiangyu He, Peisong Wang, Jian Cheng. Improving Extreme Low-bit Quantization with Soft Threshold. IEEE Transactions on Circuits and Systems for Video Technology (TCSVT), 2022.
  33. Anda Cheng, Peisong Wang, Xi Sheryl Zhang, Jian Cheng. Differentially Private Federated Learning with Local Regularization and Sparsification. CVPR 2022.
  34. Weixiang Xu, Xiangyu He, Ke Cheng, Peisong Wang, Jian Cheng. Towards Fully Sparse Training: Information Restoration with Spatial Similarity. AAAI 2022.
  35. Anda Cheng, Jiaxing Wang, Xi Sheryl Zhang, Qiang Chen, Peisong Wang, Jian Cheng. Dpnas: Neural architecture search for deep learning with differential privacy. AAAI 2022.
  36. Zhexin Li, T Yang, Peisong Wang, Jian Cheng. Q-vit: Fully differentiable quantization for vision transformer. Arxiv (2022)
  37. Peisong Wang*, Fanrong Li*, Gang Li, Jian Cheng. Extremely Sparse Networks via Binary Augmented Pruning for Fast Image Classification. IEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2021.
  38. Weihan Chen, Peisong Wang, Jian Cheng. Towards mixed-precision quantization of neural networks via constrained optimization. ICCV 2021.
  39. Zhexin Li, Peisong Wang, Z Wang, Jian Cheng. Fixed-point quantization for vision transformer. China Automation Congress (CAC),2021.
  40. Weixiang Xu, Qiang Chen, Xiangyu He, Peisong Wang, Jian Cheng. Improving binary neural networks through fully utilizing latent weights. Arxiv 2021.
  41. Zhenmeng Zuo, Zhexin Li, Peisong Wang, Weihan Chen, Jian Cheng. Towards Binarized MobileNet via Structured Sparsity. ICIG 2021.
  42. Zili Liu, Peisong Wang, Zaixing Li. More-similar-less-important: Filter pruning via kmeans clustering. ICME 2021.
  43. Xiangyu He, Jiahao Lu, Weixiang Xu, Qinghao Hu, Peisong Wang, Jian Cheng. Generative zero-shot network quantization. CVPR 2021 workshop.
  44. Peisong Wang, Xiangyu He, Qiang Chen, Anda Cheng, Qingshan Liu and Jian Cheng. Unsupervised Network Quantization via Fixed-point Factorization. IEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2020.
  45. Qiang Chen, Anda Cheng, Xiangyu He, Peisong Wang, Jian Cheng. Spatialflow: Bridging all tasks for panoptic      segmentation. IEEE Transactions on Circuits and Systems for Video Technology (TCSVT), 2020.
  46. Qiang Chen, Peisong Wang, Anda Cheng, Wanguo Wang, Yifan Zhang, Jian Cheng. Robust one-stage object detection with      location-aware classifiers. Pattern Recognition (PR), 2020.
  47. Peisong Wang, Qiang Chen, Xiangyu He and Jian Cheng. Towards Accurate Post-training Network Quantization via Bit-Split and Stitching. ICML 2020.
  48. Peisong Wang*, Xiangyu He*, Gang Li, Tianli Zhao and Jian Cheng. Sparsity-Inducing Binarized Neural Networks. AAAI 2020.
  49. Gang Li, Peisong Wang, Zejian Liu, Cong Leng, Jian Cheng. Hardware acceleration of CNN with one-hot quantization of weights and activations. DATE 2020.
  50. Weixiang Xu, Xiangyu He, Tianli Zhao, Qinghao Hu, Peisong Wang, Jian Cheng. Soft Threshold Ternary Networks. IJCAI 2020.
  51. Xiangyu He, Zitao Mo, Ke Cheng, Weixiang Xu, Qinghao Hu, Peisong Wang, Qingshan Liu, Jian Cheng. Proxybnn: Learning binarized neural networks via proxy matrices. ECCV 2020.
  52. Weihan Chen, Peisong Wang, Jian Cheng. Towards Convolutional Neural Networks Compression via Global&Progressive Product Quantization. BMVC 2020.
  53. Xiangyu He, Peisong Wang, Jian Cheng. K-nearest neighbors hashing. CVPR 2019.
  54. Xiangyu He, Zitao Mo, Peisong Wang, Yang Liu, Mingyuan Yang, Jian Cheng. Ode-inspired network design for single image super-resolution. CVPR 2019. 
  55. Zhe Li, Peisong Wang, Hanqing Lu, Jian Cheng. Reading selectively via Binary Input Gated Recurrent Unit. IJCAI 2019.
  56. Xiangyu He, Zitao Mo, Qiang Chen, Anda Cheng, Peisong Wang, Jian Cheng. Location-aware upsampling for semantic segmentation. Arxiv 2019.
  57. Xiangyu He, Ke Cheng, Qiang Chen, Qinghao Hu, Peisong Wang, Jian Cheng. Compact global descriptor for neural networks. Arxiv 2019.
  58. Peisong Wang, Qinghao Hu, Zhiwei Fang, Chaoyang Zhao and Jian Cheng. DeepSearch: A fast image search framework for mobile devices.  ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM), 2018.
  59. Jian Cheng, Peisong Wang, Gang Li, Qing-hao Hu, Hanqing Lu. Recent advances in efficient computation of deep  convolutional neural networks. Frontiers of Information Technology & Electronic Engineering (FITEE), 2018.
  60. Peisong Wang, Qinghao Hu, Yifang Zhang, Chunjie Zhang, Yang Liu and Jian Cheng. Two-Step Quantization for Low-bit Neural Networks. CVPR 2018.
  61. Qinghao Hu, Peisong Wang, Jian Cheng. From hashing to cnns: Training binary weight networks via hashing. AAAI 2018.
  62. Qinghao Hu, Gang Li, Peisong Wang, Yifan Zhang, Jian Cheng. Training binary weight networks via semi-binary decomposition. ECCV 2018.
  63. Guibo Zhu, Jinqiao Wang, Peisong Wang, Yi Wu, Hanqing Lu. Feature distilled tracking. IEEE transactions on cybernetics (TC), 2017.
  64. Peisong Wang and Jian Cheng. Fixed-point Factorized Networks. CVPR 2017.
  65. Peisong Wang, Qiang Song, Hua Han, Jian Cheng. Sequentially supervised long short-term memory for gesture recognition. Cognitive Computation, 2016.
  66. Peisong Wang and Jian Cheng. Accelerating Convolutional Neural Networks for Mobile Applications. ACM Multimedia, 2016.

科研活动

   
科研项目

（ 1 ） 深度神经网络无监督定点量化方法研究, 负责人, 国家自然科学基金青年基金, 2020-01--2022-12

（ 2 ） 面向端侧场景的高效视觉语言模型压缩与自适应部署研究, 负责人, 国家自然科学基金面上项目, 2026-01--2029-12

（ 3 ） 基于目标行为的视频流内容安全分析与智能检索, 课题负责人, 国家自然科学基金重点项目, 2020-01--2023-12

（ 4 ） 进化与推演环境-软硬协同的底层加速库, 子课题负责人, 中科院战略先导A类项目, 2020-07--2025-06

（ 5 ） 面向通用视觉的机器学习理论与方法，子课题负责人，科技创新2030重大项目，2023-01--2025-12

（ 6） 面向大模型的异构算力融合的智能编译器研发，课题负责人，江苏省科技重大专项项目，2025-01--2027-12

（ 7 ） 深度学习模型小型化和硬件加速合作项目（四期），负责人，华为合作项目，2022-09--2023-07

（ 8 ） LLM长序列推理加速算法技术合作项目，负责人，华为合作项目，2024-06--2025-05

（ 9 ） 基于自适应低精度和混合精度计算的模型训练加速方法研究，负责人，百度松果基金，2023-11--2024-10