Archive – Song Han

July 2018: “AMC: AutoML for Model Compression and Acceleration on Mobile Devices” accepted by ECCV’18. This paper use AI to do model compression, rather than rely on human heuristics to do it. AMC can automate the model compression process, achieve better compression ratio, and also be more sample efficient. It takes shorter time can do better than rule-based heuristics.
June 2018: Song presents invited paper “Bandwidth Efficient Deep Learning” at Design Automation Conference (DAC’18). The paper talks about techniques to save memory bandwith, networking bandwidth, and engineer bandwdith for efficient deep learning.
May 2018: “Path-Level Network Transformation for Efficient Architecture Search” accepted by Internatinal Conference on Machine Learning (ICML’18).
Mar 26, 2018: Song presented Deep Gradient Compression at NVIDIA GPU Technology Conference.
Feb 26, 2018: Song presented “Bandwidth Efficient Deep Learning: Challenges and Trade-offs” at FPGA’18 panel session.
Jan 29, 2018: Deep Gradient Compression is accepted by ICLR’18. This technique can reduce the communication bandwidth by 500x and improves the scalability of large-scale distributed training. [slides].
Dec 8/9, 2017: Song presented Deep Gradient Compression: Reducing the Communication Bandwidth for Distributed Training at NIPS Workshop on Machine Learning Systems and NIPS Workshop on Deep Learning at Supercomputer Scale, Long Beach.
Dec 6, 2017: Yi and Song presented “Fast-speed Intelligent Video Analytics” at NIPS 2017 demo session, Long Beach.
Sep 1, 2017: Song finished his PhD thesis: Efficient Methods and Hardware for Deep Learning.
July 26, 2017: Song presented Exploring the Regularity of Sparse Structure in Convolutional Neural Networks at CVPR’17 TMCV workshop, Honolulu.
June 1, 2017: Song passed PhD defense. [video]
April 24, 2017: Song presented Dense-Sparse-Dense training, a regularization technique for deep neural networks, at ICLR’17, Toulon, France. [DSD model zoo] [slides]
Feb 24 2017: Song received Best Paper Award for the paper ESE: Efficient Speech Recognition Engine with Sparse LSTM on FPGA at International Symposium on Field-Programmable Gate Arrays (FPGA), Monterey, CA.
Feb 22 2017: Song presents Deep Learning – Tutorial and Recent Trends at FPGA’17, Monterey. [video]
Feb 6 2017: DSD: Dense-Sparse-Dense Training for Deep Neural Networks is accepted by International Conference on Learning Representations (ICLR) 2017.
Feb 6 2017: Trained Tenary Quantization is accepted by International Conference on Learning Representations (ICLR) 2017.
Feb 1 2017: Song presented “Efficient Methods and Hardware for Deep Learning” at Efficient Neural Network Summit, Cadence, San Jose.
Dec 12 2016: Song presented From Compression to Acceleration: Efficient Methods and Hardware for Deep Learning at MIT, Cambridge.
Dec 9 2016: Song received Best Paper Honorable Mention at NIPS’16 workshop on Efficient Methods for Deep Neural Networks, Barcelona, Spain.
Nov 20 2016: “ESE: Efficient Speech Recognition Engine with Sparse LSTM on FPGA” has been accepted to appear at FPGA’17 as a full paper, it is also selected for oral presentation at NIPS’16 workshop on Efficient Methods for Deep Neural Networks.
Oct 28 2016: Song received Best Poster Award at 2016 Stanford Cloud Workshop for his poster entiled “Deep Compression, EIE and DSD:  Deep Learning Model Compression, Acceleration, and Regularization”.
Oct 24 2016: Song presented “Deep Compression and EIE: Deep Neural Network Model Compression and Hardware Acceleration” at 2016 IBM Research Workshop on Architectures for Cognitive Computing and Datacenters, Yorktown Heights.
Sep 26 2016: Song presented “Deep Neural Network Model Compression and an Efficient Inference Engine” at O’reilly Artificial Intelligence Conference, New York.
June 20 2016: Song presented “EIE: Efficient Inference Engine on Compressed Deep Neural Network” at International Symposium on Computer Architecture, Seoul, Korea.
June 10 2016: Song presented “Deep Compression, DSD Training and EIE: deep neural network model compression, regularization and hardware acceleration” at Microsoft Research, Redmond.
May 4 2016: Song received Best Paper Award in International Conference on Learning Representations (ICLR), San Juan, Puerto Rico.