Song Han

Associate Professor, MIT EECS

Media & News

TinyML & MCUNet [NeurIPS’20 spotlight][NeurIPS’21]:
– MIT Homepage & MIT News, Learning on the edge
MIT NewsTiny machine learning design alleviates a bottleneck in memory usage on internet-of-things devices
– WiredAI Algorithms Are Slimming Down to Fit in Your Fridge 
– MIT NewsSystem brings deep learning to “internet of things” devices
– IBMNew IBM-MIT system brings AI to microcontrollers – paving the way to ‘smarter’ IoT

SpAtten [HPCA’21]:
– MIT Homepage Spotlight, A language learning system that pays attention — more efficiently than ever before

QuantumNAS [HPCA’22]:
– MIT News, Making quantum circuits more robust

DiffAugment [NeurIPS’20]:
– Venture BeatMIT researchers claim augmentation technique can train GANs with less data

Once-For-All Network [ICLR’20]:
– Venture BeatMIT aims for energy efficiency in AI model training
– MIT NewsReducing the carbon footprint of artificial intelligence
– Qualcomm, Research from MIT shows promising results for on-device AI

Hardware-Aware Transformer [ACL’20]:
– MIT News, Shrinking deep learning’s carbon footprint
– Venture Beat, New AI technique speeds up language models on edge devices

Temporal Shift Module [ICCV’19]: 
– NVIDIANew MIT Video Recognition Model Dramatically Improves Latency on Edge Devices
– MIT Technology ReviewPowerful computer vision algorithms are now small enough to run on your phone
– EngadgetMIT-IBM developed a faster way to train video recognition AI
– MIT NewsFaster video recognition for the smartphone era

ProxylessNAS [ICLR’19]:
– IEEE SpectrumUsing AI to Make Better AI
– MIT NewsKicking neural network design automation into high gear