Python递归实现神经网络算法

import numpy as np

class NeuralNetwork:
    def __init__(self, sizes):
        self.num_layers = len(sizes)
        self.sizes = sizes
        self.biases = [np.random.randn(y, 1) for y in sizes[1:]]
        self.weights = [np.random.randn(y, x) for x, y in zip(sizes[:-1], sizes[1:])]

    def feedforward(self, a):
        if self.num_layers == 2:
            return self.sigmoid(np.dot(self.weights[0], a) + self.biases[0])
        else:
            return self.sigmoid(np.dot(self.weights[0], self.feedforward(a)) + self.biases[0])

    def backprop(self, x, y):
        if self.num_layers == 2:
            delta_b, delta_w = self.backprop2(x, y)
        else:
            delta_b, delta_w = self.backprop(self.feedforward(x), y)
        return delta_b, delta_w

    def backprop2(self, x, y):
        delta_b = [np.zeros(b.shape) for b in self.biases]
        delta_w = [np.zeros(w.shape) for w in self.weights]
        z = np.dot(self.weights[0], x) + self.biases[0]
        a = self.sigmoid(z)
        delta = (a - y) * self.sigmoid_prime(z)
        delta_b[0] = delta
        delta_w[0] = np.dot(delta, x.transpose())
        return delta_b, delta_w

    def sigmoid(self, z):
        return 1.0 / (1.0 + np.exp(-z))

    def sigmoid_prime(self, z):
        return self.sigmoid(z) * (1 - self.sigmoid(z))

    def evaluate(self, test_data):
        test_results = [(np.argmax(self.feedforward(x)), y) for (x, y) in test_data]
        return sum(int(x == y) for (x, y) in test_results)

    def SGD(self, training_data, epochs, mini_batch_size, eta, test_data=None):
        n = len(training_data)
        if test_data:
            n_test = len(test_data)
        for j in range(epochs):
            np.random.shuffle(training_data)
            mini_batches = [training_data[k:k+mini_batch_size] for k in range(0, n, mini_batch_size)]
            for mini_batch in mini_batches:
                self.update_mini_batch(mini_batch, eta)
            if test_data:
                print("Epoch {0}: {1} / {2}".format(j, self.evaluate(test_data), n_test))
            else:
                print("Epoch {0} complete".format(j))

    def update_mini_batch(self, mini_batch, eta):
        delta_b = [np.zeros(b.shape) for b in self.biases]
        delta_w = [np.zeros(w.shape) for w in self.weights]
        for x, y in mini_batch:
            delta_delta_b, delta_delta_w = self.backprop(x, y)
            delta_b = [db+ddb for db, ddb in zip(delta_b, delta_delta_b)]
            delta_w = [dw+ddw for dw, ddw in zip(delta_w, delta_delta_w)]
        self.weights = [w-(eta/len(mini_batch))*nw for w, nw in zip(self.weights, delta_w)]
        self.biases = [b-(eta/len(mini_batch))*nb for b, nb in zip(self.biases, delta_b)]