摘要
针对BP神经网络存在的训练速度慢、容易陷于局部极小点的问题,提出了通过遗传算法优化神经网络的方法,利用遗传算法,将BP网络的权值和阈值按一定的顺序级联起来,形成一个实数数组,作为遗传算法的一个染色体。然后构造这样的染色体群,通过遗传算法先对网络的权值和阈值做快速的学习,以此选择出BP网络的初始权值和阈值,然后进行BP算法的训练。改进的BP神经网络算法有效克服了神经网络的缺点,提高了网络训练速度。通过训练评估实例,对优化前后的神经网络进行训练,通过仿真结果可以看到,优化后的神经网络优于未优化的神经网络。
Aiming at the problem that BP neural network is slow in training speed and easy to get stuck in local minimum point, a method of optimizing BP neural network by Genetic Algorithm is proposed, the weight and threshold of BP network are concatenated in a certain order to form a real array, which is regarded as a chromosome of genetic algorithm. Then this chromosome group is constructed, and the weights and thresholds of the network are quickly learned by genetic algorithm, so as to select the initial weights and thresholds of the BP network, and then the BP algorithm is trained. The improved BP neural network algorithm effectively overcomes the shortcomings of the neural network and improves the training speed of the network. Through the example of training evaluation, the neural network before and after optimization is trained. The simulation results show that the optimized neural network optimizes the non-optimized neural network.
出处
《计算机科学与应用》
2022年第1期262-267,共6页
Computer Science and Application