Parallel Scale Cropping Deep Convolutional Neural Network Model

HUANG You-wen; FANG Yong-ping

Chinese

您当前的位置：

首页 >

文章列表页 >

Parallel Scale Cropping Deep Convolutional Neural Network Model

更新时间：2026-09-04

- Parallel Scale Cropping Deep Convolutional Neural Network Model
- Vol. 36, Issue 4, Pages: 6-11(2019)
- 作者机构：
  
  江西理工大学机构信息工程学院,江西,赣州,341000
- 作者简介：
- 基金信息：
- DOI：
  CLC：
- Published：2019
- 稿件说明：
移动端阅览
黄友文, 方永平. 并行尺度裁切的深度卷积神经网络模型[J]. 微电子学与计算机, 2019, 36(4): 6-11.

HUANG You-wen, FANG Yong-ping. Parallel Scale Cropping Deep Convolutional Neural Network Model[J]. 2019, 36(4): 6-11.
黄友文, 方永平. 并行尺度裁切的深度卷积神经网络模型[J]. 微电子学与计算机, 2019, 36(4): 6-11. DOI：

HUANG You-wen, FANG Yong-ping. Parallel Scale Cropping Deep Convolutional Neural Network Model[J]. 2019, 36(4): 6-11. DOI：

摘要

针对并行卷积神经网络(PCNN)的参数过多

模型训练时间成本高的问题

本文提出了并行尺度裁切卷积神经网络(PSC-CNN).PSC-CNN算法是将并行卷积神经网络其中一路(Path A)的输入及该通路的特征提取层的输出通过Crop层裁切得到新的尺寸的图像供给另一路(Path B)网络作为输入.这样

Path A的输入图像在数据层经过一次随机裁剪

Path B则经过了两次裁剪操作

增加了输入数据的多样性

提升了模型的学习能力.算法以AlexNet为基础网络

分别设计了对应的PCNN及PSC-CNN模型

在数据集Caltech101、Caltech256上进行实验.实验结果表明

相较原始的PCNN

本文提出的改进算法有效的提升了分类准确度同时缩短了训练时间.

Abstract

As the excessive parameters of parallel convolutional neural network (PCNN) and the high cost of model training time

this paper proposes a parallel scale cropping convolutional neural network (PSC-CNN). The PSC-CNN algorithm is that obtains an input for one path (Path A) of the parallel convolutional neural network and an output of the feature extraction layer of the path

and then to get a new size image as input of the other path (Path B) network through the Crop layer. In this way

the input image of Path A undergoes a random cropping in the data layer

and Path B undergoes two cropping operations

which increases the data diversity and improves the learning ability. The algorithm is based on AlexNet

and the corresponding PCNN

PSC-CNN models are designed respectively. Experiments are carried out on the datasets Caltech101 and Caltech256. Experimental results show that compared with the original PCNN

the improved algorithm proposed in this paper effectively improves the classification accuracy and shortens the training time.

关键词

Keywords

references

HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 770-778.

RUMELHART D E, HINTON G E, WILLIAMS R J. Learning representations by back-propagating errors[J]. nature, 1986, 323(6088): 533. DOI: 533

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

KRIZEVSKY A, SUTSKEVER I, HINTON G E. Imagenet classification with deep convolutional neural networks[C]//Advances in neural information processing systems. 2012: 1097-1105.

HINTON G E, SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. science, 2006, 313(5786): 504-507. DOI: 504

JIA Y, SHELHAMER E, DONAHUE J, et al. Caffe: Convolutional architecture for fast feature embedding[C]//Proceedings of the 22nd ACM international conference on Multimedia. ACM, 2014: 675-678.

HE K, ZHANG X, REN S, et al. Delving deep into rectifiers: Surpassing human-level performance on imagenet classification[C]//Proceedings of the IEEE international conference on computer vision. 2015: 1026-1034.

GLOROT X, BENGIO Y. Understanding the difficulty of training deep feedforward neural networks[C]//Proceedings of the thirteenth international conference on artificial intelligence and statistics. 2010: 249-256.

GUO L, LI F, LIEW A W C.Image esthetic evaluation using parallel deep convolution neural network[C]//Digital image computing: techniques and applications(DICTA), 2016 international conference on.IEEE, 2016: 1-5.

汤鹏杰, 王瀚漓, 左凌轩.并行交叉的深度卷积神经网络模型[J].中国图象图形学报, 2016, 21(3):339-347. DOI: 339

杨远飞, 曾上游, 冯燕燕, 等.基于并行和切片的深度卷积网络设计研究[J].微电子学与计算机, 2018, 35(3):120-124. DOI: 120

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Research on the design of blockchain-based AI-generated content traceability framework

Implementation and verification of a dual-chain power gating strategy for low power design

Simulation on the DC characteristics of double heterojunction p-GaN HEMT with AlN/GaN hybrid buffer layer

A high-efficiency charge pump for ultra-low voltage boost systems

Design of ultra-low power real-time clock circuits

Related Author

No data

Related Institution

No data

AI问答

Postal code：100079
Tel：（010）53879206 Email：tmw@bjxintong.com.cn
Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09082226号-64 京公网安备11010602201714号
It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.

⁰