基于粒子群算法优化卷积神经网络结构

Optimization of Convolutional Neural Network Structure Based on Particle Swarm Optimization

目前卷积神经网络(Convolutional Neural Network,CNN)在各领域应用非常广泛.训练的神经网络分类器的性能取决于其结构,而寻找一个最优CNN结构非常困难,故现有的方法多依赖于经验.提出的IPSOCNN(Improved Particle Swarm Optimization Convolutional NeuralNetwork)算法,利用粒子群(Particle Swarm Optimization,PSO)算法搜索范围广、收敛速度快的特点,自动选择CNN的最佳结构,解决人工选择的不确定性问题.实验表明,相比人工选择的最佳CNN结构及现有优化CNN算法,IPSOCNN算法的结果正确率更高.

At present,the Convolutional Neural Network(CNN)is widely used in various fields.The performance of the trained neural network classifier depends on its structure,and it is very difficult to find an optimal CNN structure,so the existing methods rely on experience.The proposed IPSOCNN(Improved Particle Swarm Optimization Convolutional Neural Network)algorithm uses the Particle Swarm Optimization(PSO)algorithm,which has a wide search range and fast convergence speed,to select the optimal structure of CNN automatically and to solve the problem of manual selection uncertainty.Experiments show that the IPSOCNN algorithm has a higher correct rate than that of the optimal CNN structure and the existing optimized CNN algorithm.