一种基于模拟退火算法改进的卷积神经网络

满凤环; 陈秀宏; 何佳佳

Chinese

您当前的位置：

首页 >

文章列表页 >

一种基于模拟退火算法改进的卷积神经网络

更新时间：2026-09-04

- 一种基于模拟退火算法改进的卷积神经网络
- An Improved Convolutional Neural Network Based on Simulated Annealing Algorithm
- 微电子学与计算机 2017年34卷第9期页码：58-62
- 作者机构：
  
  江南大学数字媒体学院,江苏,无锡,214122
- 作者简介：
- 基金信息：
  
  国家自然科学基金 61373055江苏省2015年度普通高校研究生科研创新计划项目 KYLX15_1191
- DOI：
  中图分类号：
- 纸质出版：2017
- 稿件说明：
移动端阅览
满凤环, 陈秀宏, 何佳佳. 一种基于模拟退火算法改进的卷积神经网络[J]. 微电子学与计算机, 2017, 34(9): 58-62.

MAN Feng-huan, CHEN Xiu-hong, HE Jia-jia. An Improved Convolutional Neural Network Based on Simulated Annealing Algorithm[J]. 2017, 34(9): 58-62.
满凤环, 陈秀宏, 何佳佳. 一种基于模拟退火算法改进的卷积神经网络[J]. 微电子学与计算机, 2017, 34(9): 58-62. DOI：

MAN Feng-huan, CHEN Xiu-hong, HE Jia-jia. An Improved Convolutional Neural Network Based on Simulated Annealing Algorithm[J]. 2017, 34(9): 58-62. DOI：

摘要

卷积神经网络(CNN) 是一种深度学习技术

其目标是构建一个接近人类的智能来执行任何知识的任务.使用基于交叉熵损失函数的模拟退火算法对dropout正则化

即有些单元值被抑制的CNN进行优化训练

而在测试阶段所使用的模型平均方法则同时考虑了保留概率和池化区域内单元值所占概率.在MNIST手写数据库和CMU-PIE的部分图像库的实验结果表明

在同一CNN结构和相同迭代次数时

此方法均优于其他方法

不仅能够得到更好的识别率而且更能防止过拟合的问题.

Abstract

CNN is a deep learning method

where its aim is to the goal of artificial intelligence that could perform any intellectual tasks.This paper proposes to optimize CNN with dropout regularization that it can some units became zero using simulated annealing algorithm based on cross entropy loss function

also proposes model averaging method that it combines retaining probability and probability of each unit within pooling region at test time.We do experiments on MNIST handwritten database and part of CMU-PIE database.Under the same structure and the same number of iterations

the method is superior to other methods.It can get a better recognition rate and can be better to overcome the over-fitting problem.

关键词

Keywords

references

Rere L M, Fanany M I, Arymurthy A M. Metaheuristic algorithms for convolution neural network[J]. Computational Intelligence and Neuroscience, 2016(1): 1-13. DOI: 1

Kirkpatrick S. Optimization by simulated annealing: Quantitative studies[J]. Journal of statistical physics, 1984, 34(5/6): 975-986. DOI: 975

Nair V, Hinton G E. Rectified linear units improve restricted boltzmann machines[C]//Proceedings of the27th International Conference on Machine Learning (ICML-10). IEEE, 2010: 807-814.

Bouvrie J. Notes on convolutional neural networks[R]. MIT CBCL Tech Report, 2006: 38-44.

Le Cun Y, Kavukcuoglu K, Farabet C. Convolutional networks and applications in vision[C]//ISCAS. IEEE, 2010: 253-256.

Zhining Y, Yunming P. The genetic convolutional neural network model based on random sample[J]. International Journal of u-and e-Service, Science and Technology, 2015, 8(11): 317-326. DOI: 317

Rosa G, Papa J, Marana A, et al. Fine-tuning convolutional neural networks using harmony search[C]//Iberoamerican Congress on Pattern Recognition. Springer International Publishing. ACM, 2015: 683-690.

Wu H, Gu X. Towards dropout training for convolutional neural networks[J]. Neural Networks, 2015, 71(2015): 1-10. DOI: 1

Zeiler M D, Fergus R. Stochastic pooling for regularization of deep convolutional neural networks[J]. arXiv preprint, 2013.

Vedaldi A, Lenc K. Matconvnet: Convolutional neural networks for matlab[C]//Proceedings of the 23rd ACM international conference on Multimedia. ACM, 2015: 689-692.

El-Ghazali T. Metaheuristics: from design to implementation[J]. onh Wiley and Sons Inc. Chichester, 2009(9): 10-11. DOI: 10

Krizhevsky A, Sutskever I, Hinton G E. Imagenet classification with deep convolutional neural networks[C]//Advances in neural information processing systems. IEEE, 2012: 1097-1105.

Niu X X, Suen C Y. A novel hybrid CNN-SVM classifier for recognizing handwritten digits[J]. Pattern Recognition, 2012, 45(4): 1318-1325. DOI: 1318

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

反熔丝FPGA线长驱动布局算法

基于改进RT-DETR的航拍小目标检测算法

一种用于极低压boost系统的高效率电荷泵

超低功耗实时时钟电路的设计

一种基于特征上下文理解与切片分析的肺结核检测算法