手动搭建神经网络应用于手写数字识别

import numpy as np
import scipy.special
import matplotlib.pyplot as plt
from PIL import Image

class neuralNetwork :

    # 用于神经网络初始化
    def __init__(self, inputnodes, hiddennodes, outputnodes, learningrate):
        # 输入层节点数
        self.inodes = inputnodes
        # 隐层节点数
        self.hnodes = hiddennodes
        # 输出层节点数
        self.onodes = outputnodes
        # 学习率
        self.lr = learningrate

        # 初始化输入层与隐层之间的权重
        self.wih = np.random.normal(0.0, pow(self.hnodes, -0.5), (self.hnodes, self.inodes))
        # 初始化隐层与输出层之间的权重
        self.who = np.random.normal(0.0, pow(self.onodes, -0.5), (self.onodes, self.hnodes))

        # 激活函数（S函数）
        self.activation_function = lambda x: scipy.special.expit(x)
        # 逆激活函数
        self.iactivation_function = lambda y: scipy.special.logit(y)

    # 设置权重
    def setweights(self, wih, who):
        self.wih = wih
        self.who = who

    # 神经网络学习训练
    def train(self, inputs_list, targets_list):
        # 将输入数据转化成二维矩阵
        inputs = np.array(inputs_list, ndmin=2).T
        # 将输入标签转化成二维矩阵
        targets = np.array(targets_list, ndmin=2).T

        # 计算隐层的输入
        hidden_inputs = np.dot(self.wih, inputs)
        # 计算隐层的输出
        hidden_outputs = self.activation_function(hidden_inputs)

        # 计算输出层的输入
        final_inputs = np.dot(self.who, hidden_outputs)
        # 计算输出层的输出
        final_outputs = self.activation_function(final_inputs)

        # 计算输出层误差
        output_errors = targets - final_outputs
        # 计算隐层误差
        hidden_errors = np.dot(self.who.T, output_errors)

        # 更新隐层与输出层之间的权重
        self.who += self.lr * np.dot((output_errors * final_outputs * (1.0 - final_outputs)), np.transpose(hidden_outputs))
        # 更新隐层与输出层之间的权重
        self.wih += self.lr * np.