bpnnet_156

  在实际应用中，原始的BP算法很难胜任，因此出现了很多的改进算法。BP算法的改进主要有两种途径，一种是采用启发式学习方法，另一种则是采用更有效的优化算法。本例采用动量BP算法，来实现对网络的训练过程，动量法降低了网络对于误差曲面局部细节的敏感性，有效地抑制网络陷于局部极小。 (In practical applications, the original BP algorithm very difficult to do, so there were a lot of the improved algorithm. BP Algorithm There are two main channels, a heuristic learning method is adopted, another is the introduction of more efficient optimization algorithm. In this case the use of momentum BP algorithm to realize the network training process, momentum method to reduce the network error surface for the local details of the sensitivity of the network effectively inhibited in a local minimum.)

bpnnet_156.m
说明.txt