Shengyu Fan

Shengyu Fan

Auto-generated profile, based on publicly available information

Shengyu Fan

State Key Laboratory of Cyberspace Security Defense, Institute of Information Engineering, CAS, China

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
20 June 2025
Verified
Conference Paper
Open access
DJDian Jiao
Dian Jiao
XDXianglong Deng
Xianglong Deng
ZWZhiwei Wang
Zhiwei Wang
SFShengyu Fan
Shengyu Fan
YCYi Chen
Yi Chen
MZMingzhe Zhang
Mingzhe Zhang

19 July 2023 in ACM Transactions on Architecture and Code Optimization
Verified
Journal Article
Open access
WXWeizhi Xu
Weizhi Xu
YSYintai Sun
Yintai Sun
SFShengyu Fan
Shengyu Fan
HYHui Yu
Hui Yu
XFXin Fu
Xin Fu
The convolutional neural network (CNN) is an important deep learning method, which is widely used in many fields. However, it is very time consuming to implement the CNN where convolution usually takes most of the time. There are many zero values in feature maps and filters, which leads to redundant calculations and memory accesses if dense methods are used to compute convolution. Many works recently have made use of sparsity to skip the calculations for zero values to reduce the inference time of the CNN. On the graphics processing unit platform, current works cannot fully exploit the sparsity of the feature map and achieve satisfactory performance. Therefore, we design a new parallel strategy to transform the feature map into a new storage format to avoid the redundant computation of zero values on graphics processing units. Also considering the sparsity in the feature map, we propose a fused storage format to combine the convolution operation with the following pooling operation, to further improve the performance. We carry out experiments with mainstream CNN models and achieve better performance compared with cuDNN and cuSPARSE. For VGG-19, ResNet-50, DenseNet-121, and RegNetX-16GF, 1.97×, 2.23×, 2.74×, and 1.58× speedups respectively are obtained over cuDNN. The speedups over cuSPARSE respectively are 2.10×, 1.83×, 2.35×, and 1.35× when only using the first method.

20 June 2025
Verified
Conference Paper
Open access
SFShengyu Fan
Shengyu Fan
XDXianglong Deng
Xianglong Deng
LKLiang Kong
Liang Kong
GSGuiming Shi
Guiming Shi
GFGuang Fan
Guang Fan
MZMingzhe Zhang
Mingzhe Zhang

17 October 2025
Verified
Conference Paper
Open access
LKLiang Kong
Liang Kong
SFShengyu Fan
Shengyu Fan
XDXianglong Deng
Xianglong Deng
LCLei Chen
Lei Chen
GFGuang Fan
Guang Fan
MZMingzhe Zhang
Mingzhe Zhang

19 November 2025
Verified
Conference Paper
Open access
GSGuiming Shi
Guiming Shi
YWYuchen Wei
Yuchen Wei
SFShengyu Fan
Shengyu Fan
XDXianglong Deng
Xianglong Deng
LKLiang Kong
Liang Kong
KMKaisheng Ma
Kaisheng Ma

28 May 2025
Verified
Conference Paper
Open access
ZYZihao Yang
Zihao Yang
MZMengxin Zheng
Mengxin Zheng
SFShengyu Fan
Shengyu Fan
QLQian Lou
Qian Lou
RHRui Hou
Rui Hou
MZMingzhe Zhang
Mingzhe Zhang