A Tile-Based Feature Map Compression Method for Multi-Stride Winograd Algorithm

Ge Li

doi:10.1109/tcsii.2026.3693787

2026-05-15

PRE

DOI：10.1109/tcsii.2026.3693787

原文链接

原文求助

摘要

En 中文

基于Tile的基于Winograd的特征图压缩方法（TFCM）被提出以加速滤波器大小为3、步长为2（即(3,2)）和(3,1)的卷积运算。此外，还基于一些流行的卷积神经网络（CNN）讨论了TFCM的可行性。同时，设计了一个基于Winograd的加速单元（WAU）来实现TFCM。评估结果表明，当应用于VGG16和MobileNetV1时，所提出的方法可实现45.93%~75.28%的乘法运算节省。此外，该加速器在部署VGG16时可达到1.71 TOPS，与最先进的加速器相比，每MHz的DSP效率提高了1.27倍至2倍。

Keyword:

CNNs

winograd

feature map compression

期刊

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II-EXPRESS BRIEFS

IF:

4.9

论文数:

被引数:

机构

xi'an jiaotong university

学者数:

9.3W

论文数: 6.7W

被引数: 75

引用论文

RED: A ReRAM-based Deconvolution Accelerator

2019 Design, Automation & Test in Europe Conference & Exhibition (DATE)

IF0

2019-03-01

OAAI

Zichen Fan; Ziru Li; Bing Li; Yiran Chen; Hai Li

Accelerating Convolutional Neural Network With FFT on Embedded Hardware

IEEE Transactions on Very Large Scale Integration (VLSI) Systems

IF0

2018-09-01

OAAI

Tahmid Abtahi; Colin Shea; Amey Kulkarni; Tinoosh Mohsenin

A Post-Quantum Encryption Mechanism Based on Convolutional Neural Network Accelerator

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II-EXPRESS BRIEFS

IF4.9

2024-08-01

PREAI

Huang, Yukang; Fan, Gaopeng; Mai, Junyi; Jiang, Wanling; Hu, Jinyang; Yao, Enyi

Computing Receptive Fields of Convolutional Neural Networks

Distill

IF0

2019-11-04

OAAI

André Araujo; Wade Norris; Jack Sim

Fast Algorithms for Convolutional Neural Networks

2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

IF0

2016-06-01

OAAI

Andrew Lavin; Scott Gray

WinoTrain: Winograd-Aware Training for Accurate Full 8-bit Convolution Acceleration WinoTrain：用于精确全8位卷积加速的Winograd感知训练

2023 60th ACM/IEEE Design Automation Conference (DAC)

IF0

2023-07-09

PREAI

Mori,Pierpaolo; Sampath,Shambhavi Balamuthu; Frickenstein,Lukas; Vemparala,Manoj-Rohit; Fasfous,Nael; Frickenstein,Alexander; Stechele,Walter; Passerone,Claudio

Flexible and Efficient Convolutional Acceleration on Unified Hardware Using the Two-Stage Splitting Method and Layer-Adaptive Allocation of 1-D/2-D Winograd Units 基于两阶段分割方法和层自适应分配1-D/2-D Winograd单元的统一硬件上灵活高效的卷积加速

IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems

IF0

2024-03-01

PREAI

Chen Yang; Yaoyao Yang; Yishuo Meng; Kaibo Huo; Siwei Xiang; Jianfei Wang; Li Geng

A Flexible and Efficient FPGA Accelerator for Various Large-Scale and Lightweight CNNs 适用于各种大规模和轻量级cnn的灵活高效的FPGA加速器

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I-REGULAR PAPERS

IF5.2

2022-03-01

PREAI

Wu, Xiao; Ma, Yufei; Wang, Meiqi; Wang, Zhongfeng