Back Propagation后向传播算法 Python实现
后向传播算法用来自动调节神经网络的参数,本次实现代码完全参照Back Propagation的讲解
选取sklearn.datasets的moon数据集的前250个样本进行训练,迭代10000次之后可以得到训练准确率达到87.6%
#!/bin/bash/env python
#-*- coding: utf-8
__author__ = 'ZhangJin'
import numpy as np
class BPNN:
def __init__(self, input_array, output_array, layer_node_count_list, w_step_size=0.02, b_step_size=0.02):
# layers记录每一层的输出,网络是从第0层开始算的
self.layers = []
self.input_array = input_array
self.output_array = output_array
self.layer_node_count_list = layer_node_count_list
self.w_step_size = w_step_size
self.b_s