import numpy as np import pandas as pd # import matplotlib.pyplot as plt from sklearn.model_selection import train_test_split from sklearn.preprocessing import StandardScaler, MinMaxScaler # 读取数据 f = open(r'D:\desktop\vehicle.dat', encoding='utf-8') sentimentlist = [] count = 0 for line in f: # print(line) if count <= 22: count += 1 continue s = line.strip().split(', ') sentimentlist.append(s) count += 1 f.close() vehicle_data = pd.DataFrame(sentimentlist, columns=['Compactness', 'Circularity', 'Distance_circularity', 'Radius_ratio', 'Praxis_aspect_ratio', 'Praxis_aspect_ratio', 'Max_length_aspect_ratio', 'Scatter_ratio', 'Elongatedness', 'Praxis_rectangular', 'Length_rectangular', 'Major_variance', 'Minor_variance', 'Gyration_radius', 'Major_skewness', 'Minor_skewness', 'Minor_kurtosis', 'Hollows_ratio', 'Class']) Class = vehicle_data[['Class']] del vehicle_data['Class'] vehicle_data=MinMaxScaler().fit(vehicle_data).transform(vehicle_data) # vehicle_data=StandardScaler().fit(vehicle_data).transform(vehicle_data) Class_num = Class.copy(deep=True) # 标签数值化 uniq_list = list(np.unique(Class)) for i in range(len(Class)): for j in range(4): if Class_num.iloc[i].item() == uniq_list[j]: Class_num.loc[i] = j # 对dataframe里的数据类型进行修改 Class_num = np.array(Class_num, dtype=np.intc) Class_num = pd.DataFrame(Class_num, columns=['Class']) vehicle_data = np.array(vehicle_data, dtype=np.float) vehicle_data = pd.DataFrame(vehicle_data, columns=['Compactness', 'Circularity', 'Distance_circularity', 'Radius_ratio', 'Praxis_aspect_ratio', 'Praxis_aspect_ratio', 'Max_length_aspect_ratio', 'Scatter_ratio', 'Elongatedness', 'Praxis_rectangular', 'Length_rectangular', 'Major_variance', 'Minor_variance', 'Gyration_radius', 'Major_skewness', 'Minor_skewness', 'Minor_kurtosis', 'Hollows_ratio']) X_train, X_test, y_train, y_test = train_test_split(vehicle_data, Class_num, test_size=0.2) class Layer: def __init__(self, n_input, n_output, activation=None, weights=None, bias=None): self.activation = activation self.weights = weights if weights is not None else np.random.randn(n_input, n_output) * np.sqrt(1 / n_output) self.bias = bias if bias is not None else np.random.rand(n_output) * 0.1 self.activation_output = None def forward(self, x_input): r = np.dot(x_input, self.weights) - self.bias # 向量点积，结果为output维数 self.activation_output = self.apply_activation(r) return self.activation_output def apply_activation(self, r): if self.activation is None: return r elif self.activation == 'relu': return np.maximum(r, 0) elif self.activation == 'sigmoid': # l = len(r) # y = [] # for i in range(l): # if r[i] >= 0: # y.append(1.0 / (1 + np.exp(-r[i]))) # else: # y.append(np.exp(r[i]) / (np.exp(r[i]) + 1)) # y=np.array(y) # return y x_ravel = r.ravel() # 将numpy数组展平 length = len(x_ravel) y = [] for index in range(length): if x_ravel[index] >= 0: y.append(1.0 / (1 + np.exp(-x_ravel[index]))) else: y.append(np.exp(x_ravel[index]) / (np.exp(x_ravel[index]) + 1)) return np.array(y).reshape(r.shape) # TODO:sigmoid函数的优化 # return 1 / 1 + np.exp(-r) def apply_activation_derivative(self, r): if self.activation is None: return np.ones_like(r) elif self.activation == 'relu': grad = np.array(r, copy=True) grad[r > 0] = 1. grad[r <= 0] = 0. return grad elif self.activation == 'sigmoid': return r * (1 - r) return r class Network: def __init__(self): self.layers = [] def add_layer(self, layer): self.layers.append(layer) def feed_forward(self, x_input): # 前向传播 for layer in self.layers: x_input = layer.forward(x_input) return x_input def backward(self, X, y, learning_rate): # 反向传播 output = self.feed_forward(X) g = output * (1 - output) * (y - output) # g.size=[n_output,1] for i in reversed(range(len(self.layers))): layer = self.layers[i] if layer == self.layers[-1]: # 输出层 last_layer = self.layers[i - 1] # print(len(last_layer.activation_output)) delta_weight = [[] for k in range(len(last_layer.activation_output))] for h in range(len(last_layer.activation_output)): for j in range(len(layer.activation_output)): delta_weight[h].append(learning_rate * g[j] * last_layer.activation_output[h]) delta_weight = np.array(delta_weight) # delta_weight=learning_rate*g*last_layer.activation_output layer.weights = layer.weights + delta_weight delta_bias = -learning_rate * g layer.bias = layer.bias + delta_bias else: next_layer = self.layers[i + 1] if i + 1 == len(self.layers) - 1: # 输出层前一隐藏层 last_layer = self.layers[i - 1] delta_weight = [[] for k in range(len(last_layer.activation_output))] for k in range(len(last_layer.activation_output)): for h in range(len(layer.activation_output)): sum = 0 for j in range(len(next_layer.activation_output)): sum += next_layer.weights[h][j] * g[j] * layer.activation_output[h] * ( 1 - layer.activation_output[h]) * last_layer.activation_output[i] delta_weight[k].append(sum) delta_weight = np.array(delta_weight) layer.weights += delta_weight * learning_rate delta_bias = [] for h in range(len(layer.activation_output)): sum = 0 for j in range(len(next_layer.activation_output)): sum += next_layer.weights[h][j] * g[j] * layer.activation_output[h] * ( 1 - layer.activation_output[h]) * (-1) delta_bias.append(sum) delta_bias = np.array(delta_bias) layer.bias += delta_bias # last_layer = self.layers[i - 1] # e = layer.activation_output * (1 - layer.activation_output) * np.sum(next_layer.weights * g, axis=1) # delta_weight = [[] for k in range(len(last_layer.activation_output))] # for j in range(len(last_layer.activation_output)): # for h in range(len(layer.activation_output)): # delta_weight[j].append(learning_rate * e[h] * last_layer.activation_output[j]) # delta_weight = np.array(delta_weight) # layer.weights = layer.weights + delta_weight # delta_bias = -learning_rate * e # layer.bias = layer.bias + delta_bias # elif i == 0: # 输入层后隐藏层 else: output_layer = self.layers[-1] # TODO:隐藏层1的权重更新也要变化 # e = layer.activation_output * (1 - layer.activation_output) * np.sum(next_layer.weights * g, axis=1) # delta_weight = [[] for k in range(len(X))] # for j in range(len(X)): # for h in range(len(layer.activation_output)): # delta_weight[j].append(learning_rate * e[h] * X[j]) # delta_weight = np.array(delta_weight) # layer.weights = layer.weights + delta_weight # delta_bias = -learning_rate * e # layer.bias = layer.bias + delta_bias delta_weight = [[] for i in range(len(X))] for t in range(len(X)): for k in range(len(layer.activation_output)): sum = 0 for h in range(len(next_layer.activation_output)): for j in range(len(y)): sum += output_layer.weights[h][j] * g[j] * next_layer.activation_output[h] * ( 1 - next_layer.activation_output[h]) * (next_layer.weights[k][h]) * \ layer.activation_output[k] * (1 - layer.activation_output[k]) * X[t] delta_weight[t].append(sum) delta_weight = np.array(delta_weight)*learning_rate layer.weights += delta_weight delta_bias = [] for k in range(len(layer.activation_output)): sum = 0 for h in range(len(next_layer.activation_output)): for j in range(len(y)): sum += output_layer.weights[h][j] * g[j] * next_layer.activation_output[h] * ( 1 - next_layer.activation_output[h]) * (next_layer.weights[k][h]) * \ layer.activation_output[k] * (1 - layer.activation_output[k]) * (-1) delta_bias.append(sum) delta_bias = np.array(delta_bias)*learning_rate layer.bias = layer.bias + delta_bias # else: # last_layer = self.layers[i - 1] # delta_weight = [[] for k in range(len(last_layer.activation_output))] # for k in range(len(last_layer.activation_output)): # for h in range(len(layer.activation_output)): # sum = 0 # for j in range(len(next_layer.activation_output)): # sum += next_layer.weights[h][j] * g[j] * layer.activation_output[h] * ( # 1 - layer.activation_output[h]) * last_layer.activation_output[i] # delta_weight[k].append(sum) # delta_weight = np.array(delta_weight) # layer.weights += delta_weight * learning_rate # delta_bias = [] # for h in range(len(layer.activation_output)): # sum = 0 # for j in range(len(next_layer.activation_output)): # sum += next_layer.weights[h][j] * g[j] * layer.activation_output[h] * ( # 1 - layer.activation_output[h]) * (-1) # delta_bias.append(sum) # delta_bias = np.array(delta_bias) # layer.bias += delta_bias # print(layer.weights) def train(self, X_train, X_test, y_train, y_test, learning_rate, max_epochs): # 对标签进行one-hot编码 y_onehot = np.zeros((y_train.shape[0], 4)) y_onehot[np.arange(y_train.shape[0]), np.array(y_train).flatten()] = 1 mses = [] for epoch in range(max_epochs): for x in range(len(X_train)): self.backward(np.array(X_train.iloc[x]), y_onehot[x], learning_rate) mse = np.mean(np.square(y_onehot - self.feed_forward(X_train))) mses.append(mse) print('Epoch: #%s, MSE: %f, Accuracy: %f%%' % ( epoch + 1, float(mses[epoch]), self.accuracy(self.predict(X_test), np.array(y_test).flatten()) * 100)) def accuracy(self, y_predict, y_test): # 计算准确度 return np.sum(y_predict == y_test) / len(y_test) def predict(self, X_predict): y_predict = self.feed_forward(X_predict) y_predict = np.argmax(y_predict, axis=1) return y_predict nn = Network() # 实例化网络类 nn.add_layer(Layer(18, 4, 'relu')) # 隐藏层 1, 2=>25 nn.add_layer(Layer(4, 4, 'relu')) # 隐藏层 2, 25=>50 nn.add_layer(Layer(4, 4, 'sigmoid')) # 输出层, 25=>2 nn.train(X_train, X_test, y_train, y_test, learning_rate=3e-4, max_epochs=50) # TODO:k折交叉验证