S2Engine: A Novel Systolic Architecture for Sparse Convolutional Neural Networks

2 code implementations • 15 Jun 2021 • Jianlei Yang, Wenzhi Fu, Xingzhou Cheng, Xucheng Ye, Pengcheng Dai, Weisheng Zhao

Convolutional neural networks (CNNs) have achieved great success in performing cognitive tasks.

Optimizing Memory Efficiency of Graph Neural Networks on Edge Computing Platforms

1 code implementation • 7 Apr 2021 • Ao Zhou, Jianlei Yang, Yeqi Gao, Tong Qiao, Yingjie Qi, Xiaoyi Wang, Yunli Chen, Pengcheng Dai, Weisheng Zhao, Chunming Hu

Graph neural networks (GNN) have achieved state-of-the-art performance on various industrial tasks.

Edge-computing

SparseTrain: Exploiting Dataflow Sparsity for Efficient Convolutional Neural Networks Training

no code implementations • 21 Jul 2020 • Pengcheng Dai, Jianlei Yang, Xucheng Ye, Xingzhou Cheng, Junyu Luo, Linghao Song, Yiran Chen, Weisheng Zhao

In this paper, \textit{SparseTrain} is proposed to accelerate CNN training by fully exploiting the sparsity.

Accelerating CNN Training by Pruning Activation Gradients

no code implementations • ECCV 2020 • Xucheng Ye, Pengcheng Dai, Junyu Luo, Xin Guo, Yingjie Qi, Jianlei Yang, Yiran Chen

Sparsification is an efficient approach to accelerate CNN inference, but it is challenging to take advantage of sparsity in training procedure because the involved gradients are dynamically changed.