一种双向脉动数据流的全卷积神经网络加速器

涂凯杰; 朱宜炳; 许达文

Chinese

您当前的位置：

首页 >

文章列表页 >

一种双向脉动数据流的全卷积神经网络加速器

更新时间：2026-09-04

- 一种双向脉动数据流的全卷积神经网络加速器
- Fully convolutional neural networks accelerator for Bi-direction systolic data flow
- 微电子学与计算机 2020年37卷第1期页码：33-37
- 作者机构：
  
  1. 合肥工业大学电子科学与应用物理学院,合肥,230009
  2. 江西科技学院机构信息工程学院,江西,南昌,330098
- 作者简介：
- 基金信息：
  
  国家自然科学基金 61874124
- DOI：
  中图分类号：
- 纸质出版：2020
- 稿件说明：
移动端阅览
涂凯杰, 朱宜炳, 许达文. 一种双向脉动数据流的全卷积神经网络加速器[J]. 微电子学与计算机, 2020, 37(1): 33-37.

TU Kai-jie, ZHU Yi-bing, XU Da-wen. Fully convolutional neural networks accelerator for Bi-direction systolic data flow[J]. 2020, 37(1): 33-37.
涂凯杰, 朱宜炳, 许达文. 一种双向脉动数据流的全卷积神经网络加速器[J]. 微电子学与计算机, 2020, 37(1): 33-37. DOI：

TU Kai-jie, ZHU Yi-bing, XU Da-wen. Fully convolutional neural networks accelerator for Bi-direction systolic data flow[J]. 2020, 37(1): 33-37. DOI：

摘要

全卷积神经网络近年来被应用于深度学习中的多个领域，其不仅能处理简单的图像分类任务，还能应用于例如物体检测、语义/图像分割以及基于生成式对抗网络的生成型任务.典型的全卷积神经网络中不仅包括了传统的卷积层，还有反卷积层，它们都是计算密集型的.现在大多数研究者大都关注卷积层的设计优化，而反卷积的加速优化很少.本文提出了一种双向脉动数据流的全卷积神经网络加速器，可以同时高效地处理普通卷积层以及反卷积层.实验中选取了多个具有代表性的全卷积神经网络模型，例如DCGAN，Cascaded-FCN等.相较于以往传统的未优化的加速方案，本文所设计的加速器平均可以达到2.8倍的加速比，并且能耗降低了46.3%.

Abstract

Fully Convolutional Neural Networks (FCN) have been applied in various domains of deep learning in recent years. It can be used not only for simple image classification tasks but also for such as object detection

semantic/image segmentation and generative tasks based on Generative Adversarial Networks (GAN). Typical FCN includes not only traditional convolutional layers but also deconvolutional layers

which are computationally intensive. Most researchers are now focusing on the design optimization of the convolutional layer

but deconvolutional layer receives less attention. This paper proposes a Fully Convolutional Neural Networks accelerator using bi-direction systolic data flow

which can efficiently process the general convolution layer and deconvolution layer simultaneously. Several representative FCN models

such as DCGAN

Cascaded-FCN

etc.

have been selected in the experiment. Compared with the traditional unoptimized acceleration scheme

the proposed approach can achieve 2.8X speed up on average

and the energy consumption is reduced by 46.3%.

关键词

Keywords

references

CHEN T, DU Z, SUN N, et al. Diannao: A small-footprint high-throughput accelerator for ubiquitous machine-learning[J]. ACM Sigplan Notices, 2014, 49(4): 269-284. DOI: 269

LONG J, SHELHAMER E, DARRELL T. Fully convolutional networks for semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. Boston, Massachusetts, 2015: 3431-3440.

LECUN Y, BENGIO Y. Convolutional networks for images, speech, and time series[J]. The handbook of brain theory and neural networks, 1995, 3361(10): 1995. DOI: 1995

BROMBERGER M, BASTIAN P, BERGEEST J P, et al. FPGA-accelerated Richardson-Lucy deconvolution for 3D image data[C]//2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI). Prague, Czech Republic, 2016: 132-135.

ZEILER M D, KRISHNAN D, TAYLOR G W, et al. Deconvolutional networks[C]// 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, San Francisco, CA, USA, 2010.

WEI X, YU C H, ZHANG P, et al. Automated systolic array architecture synthesis for high throughput CNN inference on FPGAs[C]//Proceedings of the 54th Annual Design Automation Conference.Austin, TX, USA, 2017: 29.

ZHANG X, DAS S, NEOPANE O, et al. A Design Methodology for Efficient Implementation of Deconvolutional Neural Networks on an FPGA[J]. arXiv preprint arXiv: 1705.02583, 2017.

RADFORD A, METZ L, CHINTALA S. Unsupervised representation learning with deep convolutional generative adversarial networks[J]. arXiv preprint arXiv: 1511.06434, 2015.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

大模型发展综述：从BERT到Deepseek的技术演进

注意力机制与改进密集卷积引导的信息隐藏模型

基于多尺度融合与小目标增强的航拍车辆检测算法

一种用于极低压boost系统的高效率电荷泵

AlN/GaN混合缓冲层双异质结p-GaN HEMT直流特性仿真