Research

Efficient DNN Training

Efficient DNN Training Summary Model compression has been extensively studied for light-weight inference, popular means includes network pruning, weight factorization, network quantization, and neural architecture search among many others. On the other hand, the literature on efficient training appears to be much sparser, DNN training still requires us to fully train the over-parameterized neural network.

[NeurIPS 2020] ShiftAddNet: A Hardware-Inspired Deep Network

Accepted as NeurIPS 2020 regular paper! Abstract: Multiplication (e.g., convolution) is arguably a cornerstone of modern deep neural networks (DNNs). However, intensive multiplications cause expensive resource costs that challenge DNN deployment on resource-constrained edge devices, driving several attempts for multiplication-less deep networks.

[NeurIPS 2020] ShiftAddNet: A Hardware-Inspired Deep Network

Accepted as NeurIPS 2020 regular paper! Abstract: Multiplication (e.g., convolution) is arguably a cornerstone of modern deep neural networks (DNNs). However, intensive multiplications cause expensive resource costs that challenge DNN deployment on resource-constrained edge devices, driving several attempts for multiplication-less deep networks.

[NeurIPS 2020] FracTrain: Fractionally Squeezing Bit Savings Both Temporally and Spatially for Efficient DNN Training

Accepted as NeurIPS 2020 regular paper! Abstract: Recent breakthroughs in deep neural networks (DNNs) have fueled a tremendous demand for intelligent edge devices featuring on-site learning, while the practical realization of such systems remains a challenge due to the limited resources available at the edge and the required massive training costs for state-of-the-art (SOTA) DNNs.

[NeurIPS 2020] FracTrain: Fractionally Squeezing Bit Savings Both Temporally and Spatially for Efficient DNN Training

Accepted as NeurIPS 2020 regular paper! Abstract: Recent breakthroughs in deep neural networks (DNNs) have fueled a tremendous demand for intelligent edge devices featuring on-site learning, while the practical realization of such systems remains a challenge due to the limited resources available at the edge and the required massive training costs for state-of-the-art (SOTA) DNNs.

[ECCV 2020] HALO: Hardware-Aware Learning to Optimize

Accepted as ECCV 2020 regular paper! Abstract: There has been an explosive demand for bringing machine learning (ML) powered intelligence into numerous Internet-of-Things (IoT) devices. However, the effectiveness of such intelligent functionality requires in-situ continuous model adaptation for adapting to new data and environments, while the on-device computing and energy resources are usually extremely constrained.

[ECCV 2020] HALO: Hardware-Aware Learning to Optimize

Accepted as ECCV 2020 regular paper! Abstract: There has been an explosive demand for bringing machine learning (ML) powered intelligence into numerous Internet-of-Things (IoT) devices. However, the effectiveness of such intelligent functionality requires in-situ continuous model adaptation for adapting to new data and environments, while the on-device computing and energy resources are usually extremely constrained.

DNN Training Stages Understanding

Recent works show that DNN training undergoes different stages, each stage shows different effects given a hyper-parameter setting and therefore entails detailed explaination. Below I aims to analyze and share the deep understanding of DNN training, especially from the following three perspectives: