Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox
Picture for Song Han

Song Han

VISTA 2.0: An Open, Data-driven Simulator for Multimodal Sensing and Policy Learning for Autonomous Vehicles


Nov 23, 2021
Alexander Amini, Tsun-Hsuan Wang, Igor Gilitschenski, Wilko Schwarting, Zhijian Liu, Song Han, Sertac Karaman, Daniela Rus

Figure 1 for VISTA 2.0: An Open, Data-driven Simulator for Multimodal Sensing and Policy Learning for Autonomous Vehicles
Figure 2 for VISTA 2.0: An Open, Data-driven Simulator for Multimodal Sensing and Policy Learning for Autonomous Vehicles
Figure 3 for VISTA 2.0: An Open, Data-driven Simulator for Multimodal Sensing and Policy Learning for Autonomous Vehicles
Figure 4 for VISTA 2.0: An Open, Data-driven Simulator for Multimodal Sensing and Policy Learning for Autonomous Vehicles
* First two authors contributed equally. Code and project website is available here: https://vista.csail.mit.edu 
  Access Paper or Ask Questions

MCUNetV2: Memory-Efficient Patch-based Inference for Tiny Deep Learning


Oct 28, 2021
Ji Lin, Wei-Ming Chen, Han Cai, Chuang Gan, Song Han

Figure 1 for MCUNetV2: Memory-Efficient Patch-based Inference for Tiny Deep Learning
Figure 2 for MCUNetV2: Memory-Efficient Patch-based Inference for Tiny Deep Learning
Figure 3 for MCUNetV2: Memory-Efficient Patch-based Inference for Tiny Deep Learning
Figure 4 for MCUNetV2: Memory-Efficient Patch-based Inference for Tiny Deep Learning

  Access Paper or Ask Questions

RoQNN: Noise-Aware Training for Robust Quantum Neural Networks


Oct 21, 2021
Hanrui Wang, Jiaqi Gu, Yongshan Ding, Zirui Li, Frederic T. Chong, David Z. Pan, Song Han

Figure 1 for RoQNN: Noise-Aware Training for Robust Quantum Neural Networks
Figure 2 for RoQNN: Noise-Aware Training for Robust Quantum Neural Networks
Figure 3 for RoQNN: Noise-Aware Training for Robust Quantum Neural Networks
Figure 4 for RoQNN: Noise-Aware Training for Robust Quantum Neural Networks
* 19 pages, 10 figures, open-source at https://github.com/mit-han-lab/pytorch-quantum 
  Access Paper or Ask Questions

Network Augmentation for Tiny Deep Learning


Oct 17, 2021
Han Cai, Chuang Gan, Ji Lin, Song Han

Figure 1 for Network Augmentation for Tiny Deep Learning
Figure 2 for Network Augmentation for Tiny Deep Learning
Figure 3 for Network Augmentation for Tiny Deep Learning
Figure 4 for Network Augmentation for Tiny Deep Learning

  Access Paper or Ask Questions

TSM: Temporal Shift Module for Efficient and Scalable Video Understanding on Edge Device


Sep 27, 2021
Ji Lin, Chuang Gan, Kuan Wang, Song Han

Figure 1 for TSM: Temporal Shift Module for Efficient and Scalable Video Understanding on Edge Device
Figure 2 for TSM: Temporal Shift Module for Efficient and Scalable Video Understanding on Edge Device
Figure 3 for TSM: Temporal Shift Module for Efficient and Scalable Video Understanding on Edge Device
Figure 4 for TSM: Temporal Shift Module for Efficient and Scalable Video Understanding on Edge Device
* Journal preprint of arXiv:1811.08383 (TPAMI, 2020). arXiv admin note: substantial text overlap with arXiv:1910.00932 
  Access Paper or Ask Questions

LocTex: Learning Data-Efficient Visual Representations from Localized Textual Supervision


Aug 26, 2021
Zhijian Liu, Simon Stent, Jie Li, John Gideon, Song Han

Figure 1 for LocTex: Learning Data-Efficient Visual Representations from Localized Textual Supervision
Figure 2 for LocTex: Learning Data-Efficient Visual Representations from Localized Textual Supervision
Figure 3 for LocTex: Learning Data-Efficient Visual Representations from Localized Textual Supervision
Figure 4 for LocTex: Learning Data-Efficient Visual Representations from Localized Textual Supervision
* ICCV 2021. Project page: https://loctex.mit.edu/ 
  Access Paper or Ask Questions

QuantumNAS: Noise-Adaptive Search for Robust Quantum Circuits


Aug 02, 2021
Hanrui Wang, Yongshan Ding, Jiaqi Gu, Yujun Lin, David Z. Pan, Frederic T. Chong, Song Han

Figure 1 for QuantumNAS: Noise-Adaptive Search for Robust Quantum Circuits
Figure 2 for QuantumNAS: Noise-Adaptive Search for Robust Quantum Circuits
Figure 3 for QuantumNAS: Noise-Adaptive Search for Robust Quantum Circuits
Figure 4 for QuantumNAS: Noise-Adaptive Search for Robust Quantum Circuits
* 14 pages, 22 figures. Code available at https://github.com/mit-han-lab/pytorch-quantum 
  Access Paper or Ask Questions

NAAS: Neural Accelerator Architecture Search


May 27, 2021
Yujun Lin, Mengtian Yang, Song Han

Figure 1 for NAAS: Neural Accelerator Architecture Search
Figure 2 for NAAS: Neural Accelerator Architecture Search
Figure 3 for NAAS: Neural Accelerator Architecture Search
Figure 4 for NAAS: Neural Accelerator Architecture Search
* Accepted by DAC'21 
  Access Paper or Ask Questions

Efficient and Robust LiDAR-Based End-to-End Navigation


May 20, 2021
Zhijian Liu, Alexander Amini, Sibo Zhu, Sertac Karaman, Song Han, Daniela Rus

Figure 1 for Efficient and Robust LiDAR-Based End-to-End Navigation
Figure 2 for Efficient and Robust LiDAR-Based End-to-End Navigation
Figure 3 for Efficient and Robust LiDAR-Based End-to-End Navigation
Figure 4 for Efficient and Robust LiDAR-Based End-to-End Navigation
* ICRA 2021. The first two authors contributed equally to this work. Project page: https://le2ed.mit.edu/ 
  Access Paper or Ask Questions

PatchNet -- Short-range Template Matching for Efficient Video Processing


Mar 10, 2021
Huizi Mao, Sibo Zhu, Song Han, William J. Dally

Figure 1 for PatchNet -- Short-range Template Matching for Efficient Video Processing
Figure 2 for PatchNet -- Short-range Template Matching for Efficient Video Processing
Figure 3 for PatchNet -- Short-range Template Matching for Efficient Video Processing
Figure 4 for PatchNet -- Short-range Template Matching for Efficient Video Processing

  Access Paper or Ask Questions

Anycost GANs for Interactive Image Synthesis and Editing


Mar 04, 2021
Ji Lin, Richard Zhang, Frieder Ganz, Song Han, Jun-Yan Zhu

Figure 1 for Anycost GANs for Interactive Image Synthesis and Editing
Figure 2 for Anycost GANs for Interactive Image Synthesis and Editing
Figure 3 for Anycost GANs for Interactive Image Synthesis and Editing
Figure 4 for Anycost GANs for Interactive Image Synthesis and Editing
* Accepted to CVPR 2021. The code and demo are available: https://github.com/mit-han-lab/anycost-gan 
  Access Paper or Ask Questions

SpAtten: Efficient Sparse Attention Architecture with Cascade Token and Head Pruning


Jan 04, 2021
Hanrui Wang, Zhekai Zhang, Song Han

Figure 1 for SpAtten: Efficient Sparse Attention Architecture with Cascade Token and Head Pruning
Figure 2 for SpAtten: Efficient Sparse Attention Architecture with Cascade Token and Head Pruning
Figure 3 for SpAtten: Efficient Sparse Attention Architecture with Cascade Token and Head Pruning
Figure 4 for SpAtten: Efficient Sparse Attention Architecture with Cascade Token and Head Pruning
* Published as a conference paper in HPCA 2021; 15 pages, 23 figures 
  Access Paper or Ask Questions

IOS: Inter-Operator Scheduler for CNN Acceleration


Nov 02, 2020
Yaoyao Ding, Ligeng Zhu, Zhihao Jia, Gennady Pekhimenko, Song Han

Figure 1 for IOS: Inter-Operator Scheduler for CNN Acceleration
Figure 2 for IOS: Inter-Operator Scheduler for CNN Acceleration
Figure 3 for IOS: Inter-Operator Scheduler for CNN Acceleration
Figure 4 for IOS: Inter-Operator Scheduler for CNN Acceleration

  Access Paper or Ask Questions

Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution


Aug 13, 2020
Haotian Tang, Zhijian Liu, Shengyu Zhao, Yujun Lin, Ji Lin, Hanrui Wang, Song Han

Figure 1 for Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution
Figure 2 for Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution
Figure 3 for Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution
Figure 4 for Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution
* ECCV 2020. The first two authors contributed equally to this work. Project page: http://spvnas.mit.edu/ 
  Access Paper or Ask Questions

Hardware-Centric AutoML for Mixed-Precision Quantization


Aug 11, 2020
Kuan Wang, Zhijian Liu, Yujun Lin, Ji Lin, Song Han

Figure 1 for Hardware-Centric AutoML for Mixed-Precision Quantization
Figure 2 for Hardware-Centric AutoML for Mixed-Precision Quantization
Figure 3 for Hardware-Centric AutoML for Mixed-Precision Quantization
Figure 4 for Hardware-Centric AutoML for Mixed-Precision Quantization
* International Journal of Computer Vision (IJCV), 2020 
 * Journal preprint of arXiv:1811.08886 (IJCV, 2020). The first three authors contributed equally to this work. Project page: https://hanlab.mit.edu/projects/haq/ 
  Access Paper or Ask Questions

Tiny Transfer Learning: Towards Memory-Efficient On-Device Learning


Jul 28, 2020
Han Cai, Chuang Gan, Ligeng Zhu, Song Han

Figure 1 for Tiny Transfer Learning: Towards Memory-Efficient On-Device Learning
Figure 2 for Tiny Transfer Learning: Towards Memory-Efficient On-Device Learning
Figure 3 for Tiny Transfer Learning: Towards Memory-Efficient On-Device Learning
Figure 4 for Tiny Transfer Learning: Towards Memory-Efficient On-Device Learning

  Access Paper or Ask Questions

MCUNet: Tiny Deep Learning on IoT Devices


Jul 20, 2020
Ji Lin, Wei-Ming Chen, Yujun Lin, John Cohn, Chuang Gan, Song Han

Figure 1 for MCUNet: Tiny Deep Learning on IoT Devices
Figure 2 for MCUNet: Tiny Deep Learning on IoT Devices
Figure 3 for MCUNet: Tiny Deep Learning on IoT Devices
Figure 4 for MCUNet: Tiny Deep Learning on IoT Devices
* Demo video available here: https://youtu.be/YvioBgtec4U 
  Access Paper or Ask Questions

Differentiable Augmentation for Data-Efficient GAN Training


Jun 18, 2020
Shengyu Zhao, Zhijian Liu, Ji Lin, Jun-Yan Zhu, Song Han

Figure 1 for Differentiable Augmentation for Data-Efficient GAN Training
Figure 2 for Differentiable Augmentation for Data-Efficient GAN Training
Figure 3 for Differentiable Augmentation for Data-Efficient GAN Training
Figure 4 for Differentiable Augmentation for Data-Efficient GAN Training

  Access Paper or Ask Questions

APQ: Joint Search for Network Architecture, Pruning and Quantization Policy


Jun 15, 2020
Tianzhe Wang, Kuan Wang, Han Cai, Ji Lin, Zhijian Liu, Song Han

Figure 1 for APQ: Joint Search for Network Architecture, Pruning and Quantization Policy
Figure 2 for APQ: Joint Search for Network Architecture, Pruning and Quantization Policy
Figure 3 for APQ: Joint Search for Network Architecture, Pruning and Quantization Policy
Figure 4 for APQ: Joint Search for Network Architecture, Pruning and Quantization Policy
* Accepted by CVPR 2020 
  Access Paper or Ask Questions

HAT: Hardware-Aware Transformers for Efficient Natural Language Processing


May 28, 2020
Hanrui Wang, Zhanghao Wu, Zhijian Liu, Han Cai, Ligeng Zhu, Chuang Gan, Song Han

Figure 1 for HAT: Hardware-Aware Transformers for Efficient Natural Language Processing
Figure 2 for HAT: Hardware-Aware Transformers for Efficient Natural Language Processing
Figure 3 for HAT: Hardware-Aware Transformers for Efficient Natural Language Processing
Figure 4 for HAT: Hardware-Aware Transformers for Efficient Natural Language Processing
* Accepted to ACL 2020. 14 pages, 12 figures. Code available at http://github.com/mit-han-lab/hardware-aware-transformers.git 
  Access Paper or Ask Questions

MicroNet for Efficient Language Modeling


May 16, 2020
Zhongxia Yan, Hanrui Wang, Demi Guo, Song Han

Figure 1 for MicroNet for Efficient Language Modeling
Figure 2 for MicroNet for Efficient Language Modeling
Figure 3 for MicroNet for Efficient Language Modeling
Figure 4 for MicroNet for Efficient Language Modeling
* Accepted by PMLR 
  Access Paper or Ask Questions

GCN-RL Circuit Designer: Transferable Transistor Sizing with Graph Neural Networks and Reinforcement Learning


Apr 30, 2020
Hanrui Wang, Kuan Wang, Jiacheng Yang, Linxiao Shen, Nan Sun, Hae-Seung Lee, Song Han

Figure 1 for GCN-RL Circuit Designer: Transferable Transistor Sizing with Graph Neural Networks and Reinforcement Learning
Figure 2 for GCN-RL Circuit Designer: Transferable Transistor Sizing with Graph Neural Networks and Reinforcement Learning
Figure 3 for GCN-RL Circuit Designer: Transferable Transistor Sizing with Graph Neural Networks and Reinforcement Learning
Figure 4 for GCN-RL Circuit Designer: Transferable Transistor Sizing with Graph Neural Networks and Reinforcement Learning
* Accepted to the 57th Design Automation Conference (DAC 2020); 6 pages, 8 figures 
  Access Paper or Ask Questions

Lite Transformer with Long-Short Range Attention


Apr 24, 2020
Zhanghao Wu, Zhijian Liu, Ji Lin, Yujun Lin, Song Han

Figure 1 for Lite Transformer with Long-Short Range Attention
Figure 2 for Lite Transformer with Long-Short Range Attention
Figure 3 for Lite Transformer with Long-Short Range Attention
Figure 4 for Lite Transformer with Long-Short Range Attention
* ICLR 2020. The first two authors contributed equally to this work 
  Access Paper or Ask Questions

GAN Compression: Efficient Architectures for Interactive Conditional GANs


Mar 19, 2020
Muyang Li, Ji Lin, Yaoyao Ding, Zhijian Liu, Jun-Yan Zhu, Song Han

Figure 1 for GAN Compression: Efficient Architectures for Interactive Conditional GANs
Figure 2 for GAN Compression: Efficient Architectures for Interactive Conditional GANs
Figure 3 for GAN Compression: Efficient Architectures for Interactive Conditional GANs
Figure 4 for GAN Compression: Efficient Architectures for Interactive Conditional GANs

  Access Paper or Ask Questions

Training Kinetics in 15 Minutes: Large-scale Distributed Training on Videos


Oct 01, 2019
Ji Lin, Chuang Gan, Song Han

Figure 1 for Training Kinetics in 15 Minutes: Large-scale Distributed Training on Videos
Figure 2 for Training Kinetics in 15 Minutes: Large-scale Distributed Training on Videos
Figure 3 for Training Kinetics in 15 Minutes: Large-scale Distributed Training on Videos
Figure 4 for Training Kinetics in 15 Minutes: Large-scale Distributed Training on Videos

  Access Paper or Ask Questions

Once for All: Train One Network and Specialize it for Efficient Deployment


Aug 26, 2019
Han Cai, Chuang Gan, Song Han

Figure 1 for Once for All: Train One Network and Specialize it for Efficient Deployment
Figure 2 for Once for All: Train One Network and Specialize it for Efficient Deployment
Figure 3 for Once for All: Train One Network and Specialize it for Efficient Deployment
Figure 4 for Once for All: Train One Network and Specialize it for Efficient Deployment

  Access Paper or Ask Questions