TensorFlow2.0之numpy实现反向传播

TensorFlow2.0之numpy实现反向传播

from sklearn import datasets
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
import seaborn as sns
import numpy as np


def load_dataset():
    # 采样点数
    N_SAMPLES = 2000
    # 测试数量比率
    TEST_SIZE = 0.3
    # 生成数据集
    x, y = datasets.make_moons(n_samples=N_SAMPLES, noise=0.2, random_state=100)
    # 将2000个点按照7:3分割为训练集和测试集
    x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=TEST_SIZE, random_state=42)
    return x, y, x_train, x_test, y_train, y_test
    print(x.shape, y.shape)


# 绘制数据集的分布，x为2D坐标，y为数据点的标签
def make_plot(x, y, plot_name, file_name=None, xx=None, yy=None, preds=None, dark=False):
    if dark:
        plt.style.use('dark_background')
    else:
        sns.set_style("whitegrid")
    plt.figure(figsize=(16, 12))
    axes = plt.gca()
    axes.set(xlabel="$x_1$", ylabel="$x_2$")
    plt.title(plot_name, fontsize=30)
    plt.subplots_adjust(left=0.20)
    plt.subplots_adjust(right=0.80)
    if xx is not None and yy is not None and preds is not None:
        plt.contour(xx, yy, preds.reshape(xx.shape), levels=[.5], cmap="Greys", vmin=0, xmax=.6)
    # 绘制散点图,根据标签区分颜色
    plt.scatter(x[:, 0], x[:, 1], c=y.ravel(), s=40, cmap=plt.cm.Spectral, edgecolors='none')
    plt.savefig("dataset.svg")
    plt.close()


class Layer:
    # 全连接网络层
    def __init__(self, n_input, n_neurons, activation=None, weights=None, bias=None):
        """
         :param int n_input: 输入节点数
         :param int n_neurons: 输出节点数
         :param str activation: 激活函数类型
         :param weights: 权值张量，默认类内部生成
         :param bias: 偏置，默认类内部生成
        """
        # 通过正态分布初始化网络权值，初始化非常重要，不合适的初始化将导致网络不收敛
        self.weights = weights if weights is not None else np.random.randn(n_input, n_neurons) * np.sqrt(1 / n_neurons)
        self.bias = bias if bias is not None else np.random.rand(n_neurons) * 0.1
        # 激活函数类型
        self.activation = activation
        # 激活函数的输出值
        self.last_activation = None
        # 用于计算当前层的delta变量的中间变量
        self.error = None
        # 记录当前层的delta变量，用于计算梯度
        self.delta = None

    def _apply_activation(self, r):
        # 激活函数的输出
        if self.activation is None:
            # 无激活函数，直接返回
            return r
        elif self.activation == 'relu':
            return np.maximum(r, 0)
        # tanh激活函数
        elif self.activation == 'sigmoid':
            return 1 / (1 + np.exp(-r))
        return r

    def apply_activation_derivative(self, r):
        # 计算激活函数的导数
        # 无激活函数，导数为1
        if self.activation is None:
            return np.ones_like(r)
        # ReLU函数的导数实现
        elif self.activation == 'relu':
            grad = np.array(r, copy=True)
            grad[r > 0] = 1
            grad[r <= 0] = 0
            return grad
        # tanh函数的导数实现
        elif self.activation == 'tanh':
            return 1 - r ** 2
        # Sigmoid函数的导数实现
        elif self.activation == 'sigmoid':
            return r * (1 - r)
        return r


class NeuralNetwork:
    # 神经网络模型大类：
    def __init__(self):
        # 网络层对象列表
        self._layers = []

    def add_layer(self, layer):
        # 追加网络层
        self._layers.append(layer)

    def feed_forward(self, x):
        # 前向传播
        for layer in self._layers:
            # 依次通过各个网络层
            x = layer.activate(x)
        return x

    def train(self, X_train, X_test, y_train, y_test, learning_rate, max_epochs):
        # 网络训练函数
        # one-hot 编码
        y_onehot = np.zeros((y_train.shape[0], 2))
        y_onehot[np.arange(y_train.shape[0]), y_train] = 1

        # 将One-hot 编码后的真实标签与网络的输出计算均方误差，并调用反向传播函数更新网络参数，循环迭代训练集1000 遍即可
        mses = []
        accuracys = []
        for i in range(max_epochs + 1):  # 训练1000 个epoch
            for j in range(len(X_train)):  # 一次训练一个样本
                self.backpropagation(X_train[j], y_onehot[j], learning_rate)
            if i % 10 == 0:
                # 打印出MSE Loss
                mse = np.mean(np.square(y_onehot - self.feed_forward(X_train)))
                mses.append(mse)
                accuracy = self.accuracy(self.predict(X_test), y_test.flatten())
                accuracys.append(accuracy)
                print('Epoch: #%s, MSE: %f' % (i, float(mse)))
                # 统计并打印准确率
                print('Accuracy: %.2f%%' % (accuracy * 100))
        return mses, accuracys

    def predict(self, X):
        return self.feed_forward(X)

    def accuracy(self, X, y):
        return np.sum(np.equal(np.argmax(X, axis=1), y)) / y.shape[0]

    # 反向传播算法实现
    def backpropagation(self, x, y, learning_rate):
        # 前向计算，得到输出值
        output = self.feed_forward(x)
        # 反向循环
        for i in reversed(range(len(self.layers))):
            # 得到当前层对象
            layer = self._layers[i]
            # 如果是输出层
            if layer == self._layers[-1]:  # 对于输出层
                layer.error = y - output  # 计算 2 分类任务的均方差的导数
                # 关键步骤：计算最后一层的 delta，参考输出层的梯度公式
                layer.delta = layer.error * layer.apply_activation_derivative(output)
            else:  # 如果是隐藏层
                next_layer = self._layers[i + 1]  # 得到下一层对象
                layer.error = np.dot(next_layer.weights, next_layer.delta)
                # 关键步骤：计算隐藏层的 delta，参考隐藏层的梯度公式
                layer.delta = layer.error * layer.apply_activation_derivative(layer.last_activation)
                # 循环更新权值
            for i in range(len(self._layers)):
                layer = self._layers[i]
                # o_i 为上一网络层的输出
                o_i = np.atleast_2d(x if i == 0 else self._layers[i - 1].last_activation)
                # 梯度下降算法，delta 是公式中的负数，故这里用加号
                layer.weights += layer.delta * o_i.T * learning_rate

 main():
    x, y, x_train, x_test, y_train, y_test = load_dataset()
    # 调用 make_plot 函数绘制数据的分布，其中 X 为 2D 坐标， y 为标签
    make_plot(x, y, "Classification Dataset Visualization ")
    plt.show()
    nn = NeuralNetwork()  # 实例化网络类
    nn.add_layer(Layer(2, 25, 'sigmoid'))  # 隐藏层 1, 2=>25
    nn.add_layer(Layer(25, 50, 'sigmoid'))  # 隐藏层 2, 25=>50
    nn.add_layer(Layer(50, 25, 'sigmoid'))  # 隐藏层 3, 50=>25
    nn.add_layer(Layer(25, 2, 'sigmoid'))  # 输出层, 25=>2
    mses, accuracys = nn.train(x_train, x_test, y_train, y_test, 0.01, 1000)

    x = [i for i in range(0, 101, 10)]

    # 绘制MES曲线
    plt.title("MES Loss")
    plt.plot(x, mses[:11], color='blue')
    plt.xlabel('Epoch')
    plt.ylabel('MSE')
    plt.savefig('训练误差曲线.svg')
    plt.close()

    # 绘制Accuracy曲线
    plt.title("Accuracy")
    plt.plot(x, accuracys[:11], color='blue')
    plt.xlabel('Epoch')
    plt.ylabel('Accuracy')
    plt.savefig('网络测试准确率.svg')
    plt.close()


if __name__ == '__main__':
    main()