Talks

• Self Driving by World Modelling (PDF), Oct. 2024
• Code-centric Large Language Model (PDF), Tsinghua University, May 2023
• Hardware-software co-design for Computer Vision（PDF, MNISC ），Tsinghua University, Apr. 2022
• Coevolution of Neural Network and Computer Architecture ( PDF ), Aug. 2019
• Speculations about Computer Architecture in Next Three Years ( PDF), Jan. 20, 2018
• Quantum Computing with Haskell and FPGA simulation ( PDF , GitHub ), Jan. 18, 2018
• Smart Embedded Vision with Quantized Neural Network ( PDF ), Tsinghua University, Jul. 8 2017
• Practical Methodology in Deep Learning ( PDF ), Peking University, Apr. 2017
• Neural Network Approximations ( PDF ), Yao Class, Tsinghua University, Nov. 2016
• Pointer Level Analysis ( PDF ), 2009

Selected Publications

• (ICCV'25 highlight) MeshLLM: Empowering Large Language Models to Progressively Understand and Generate 3D Mesh
• (2025) Step-audio: Unified understanding and generation in intelligent speech interaction
• (ECCV'24) Chat-edit-3d: Interactive 3d scene editing via text prompts
• (ICCAD'23) Sole: Hardware-software co-design of softmax and layernorm for efficient transformer inference ( arxiv)
• (2023) Occdepth: A depth-aware method for 3d semantic scene completion
• (ICCV'23) Occ$^2$Net: Robust Image Matching Based on 3D Occupancy Estimation for Occluded Regions
• (CVPR'23 highlight) Xiaotao Hu, et al. "DMVFN: A Dynamic Multi-Scale Voxel Flow Network for Video Prediction." ( arxiv, github)
• (CVPR'23 highlight) Shengchao Zhou, et al. "UniDistill: A Universal Cross-Modality Knowledge Distillation Framework for 3D Object Detection in Bird’s-Eye View." ( arxiv, github)
• (CVPR'23) Yun-Hao Cao, et al. "Three Guidelines You Should Know for Universally Slimmable Self-Supervised Learning." ( arxiv, github)
• (AAAI'23 Oral) Shuangkang Fang, et al. "One is All: Bridging the Gap Between Neural Radiance Fields Architectures with Progressive Volume Distillation." ( arxiv, github)
• (ECCV oral) Yun-Hao Cao, Peiqin Sun, Yechang Huang, Jianxin Wu and Shuchang Zhou: Synergistic Self-supervised and Quantization Learning. ( arxiv, github)
• (ECCV) Zhewei Huang, Tianyuan Zhang, Wen Heng, Boxin Shi, Shuchang Zhou: RIFE: Real-Time Intermediate Flow Estimation for Video Frame Interpolation. ( arxiv, github)
• (EACL) Yuekai Zhao, et al. Multi-split Reversible Transformers Can Enhance Neural Machine Translation
• (IJCAI) Yang Lin, et al. "Fq-vit: Fully quantized vision transformer without retraining". ( arxiv, github)
• (EMNLP) Yuekai Zhao, et al. Active learning approaches to enhancing neural machine translation. ( pdf )
• (CVPR) Peibin Chen, et al. Data-efficient semi-supervised learning by reliable edge mining
• (ICCV) Zhewei Huang, Wen Heng, Shuchang Zhou: Learning to Paint with Model-based Deep Reinforcement Learning. ( arxiv, github)
• (CVPR) Xinyu Zhou, Cong Yao, He Wen, Yuzhi Wang, Shuchang Zhou, Weiran He, Jiajun Liang: EAST: An Efficient and Accurate Scene Text Detector. ( arxiv)
• (BMVC) Shuchang Zhou, Taihong Xiao, Yi Yang, Dieqiao Feng, Qinyao He, Weiran He: GeneGAN: Learning Object Transfiguration and Attribute Subspace from Unpaired Data. ( arxiv , slide , GitHub )
• Shuchang Zhou, Yuxin Wu, Zekun Ni, Xinyu Zhou, He Wen, Yuheng Zou: DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients. ( arxiv , GitHub )
• (ACL) Fangtao Li et al. "Deceptive answer prediction with user preference graph"
• Shuchang Zhou: An Efficient Simulation Algorithm for Cache of Random Replacement Policy. PDF, slide , GitHub
• Open64 on MIPS: porting and enhancing Open64 for Loongson II
• More at Google Scholar (citation 7000+)

Competitions

• 1st place in the listening head generation track and 2nd place in the talking head generation track, ACM Multimedia ViCo 2022 Conversational Head Generation Challenge, report
• "Nuri" won the 1st place in NeurIPS'21 Machine Learning for Combinatorial Optimization Dual Task track. Code
• "Megvii-hzwer" won the 2nd place in NIPS'17 Learning to Run Challenge , a competition on teaching a skeleton to run as fast as possible. We proposed the Actor-Critic Ensemble (ACE) method ( PDF , Github).
• "Megvii" won the 1st place in all tracks in NIST TRAIT '16 , a competition on Text Recognition in the wild (OCR).

Research Areas

• Machine Learning and Artificial Intelligence
• Computer Architecture
• Stochastic Optimization

Academic Service