基于DEAP数据集的脑电情绪识别.rar

import numpy as np import pandas as pd import matplotlib.pyplot as plt from sklearn.preprocessing import normalize import os import tensorflow as tf os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # 消除警告 # Calculating accuracy and loss def testing(M, L, model): ''' arguments: M: testing dataset L: testing dataset label model: scikit-learn model return: void ''' output = model.predict(M[0:78080:32]) label = L[0:78080:32] k = 0 l = 0 for i in range(len(label)): k = k + (output[i] - label[i]) * (output[i] - label[i]) # 方差 # a good guess if (output[i] > 5 and label[i] > 5): l = l + 1 elif (output[i] < 5 and label[i] < 5): l = l + 1 print("l2 error:", k / len(label), "classification accuracy:", l / len(label), l, len(label)) # loading training and testing dataset file_path = r"C:\Users\95933\PycharmProjects\认知科学与基础" with open(file_path + '\data_training.npy', 'rb') as fileTrain: X = np.load(fileTrain) with open(file_path + '\label_training.npy', 'rb') as fileTrainL: Y = np.load(fileTrainL) X = normalize(X) # 归一化 Z = np.ravel(Y[:, [1]]) # 扁平化 Valence_Train = np.ravel(Y[:, [0]]) Arousal_Train = np.ravel(Y[:, [1]]) Domain_Train = np.ravel(Y[:, [2]]) Like_Train = np.ravel(Y[:, [3]]) with open(file_path + '\data_validation.npy', 'rb') as fileTrain: M = np.load(fileTrain) with open(file_path + '\label_validation.npy', 'rb') as fileTrainL: N = np.load(fileTrainL) M = normalize(M) L = np.ravel(N[:, [1]]) # arousa标签 Valence_Test = np.ravel(N[:, [0]]) Arousal_Test = np.ravel(N[:, [1]]) Domain_Test = np.ravel(N[:, [2]]) Like_Test = np.ravel(N[:, [3]]) def preprocess(y): result = [] for i in range(len(y)): if y[i] <= 5: result.append(0) else: result.append(1) result = np.array(result) return result # Pull out columns for X (data to train with) and Y (value to predict) X_training = X[0:468480:32] Y_training = Z[0:468480:32] # Pull out columns for X (data to train with) and Y (value to predict) X_testing = M[0:78080:32] Y_testing = L[0:78080:32] # DO Scale both the training inputs and outputs X_scaled_training = pd.DataFrame(data=X_training).values Y_scaled_training = pd.DataFrame(data=Y_training).values # It's very important that the training and test data are scaled with the same scaler. X_scaled_testing = pd.DataFrame(data=X_testing).values Y_scaled_testing = pd.DataFrame(data=Y_testing).values # 训练集标签处理 Y_scaled_training = tf.reshape(Y_scaled_training, [-1]) Y_scaled_training = preprocess(Y_scaled_training) Y_scaled_training = tf.cast(Y_scaled_training, dtype=tf.int32) Y_scaled_training = tf.one_hot(Y_scaled_training, 2) # 测试机标签处理 Y_scaled_testing = tf.reshape(Y_scaled_testing, [-1]) Y_scaled_testing = preprocess(Y_scaled_testing) Y_scaled_testing = tf.cast(Y_scaled_testing, dtype=tf.int32) Y_scaled_testing = tf.one_hot(Y_scaled_testing, 2) # print(tf.shape(X_scaled_training), np.array(X_scaled_training).shape) # print(tf.shape(Y_scaled_training), np.array(Y_scaled_training).shape) # print(Y_scaled_training[:100]) model = tf.keras.Sequential([ tf.keras.layers.Dense(70), # 输入层 tf.keras.layers.Dense(512, activation=tf.nn.relu), # 第一层 tf.keras.layers.Dropout(0.2), tf.keras.layers.Dense(1024, activation=tf.nn.relu), # 第二层 tf.keras.layers.Dropout(0.2), tf.keras.layers.Dense(1024, activation=tf.nn.relu), # 第三层 tf.keras.layers.Dense(512, activation=tf.nn.relu), # 第四层 # 输出层，2分类问题，激活函数为softmax tf.keras.layers.Dense(2, activation=tf.nn.softmax) # 输出层 ]) # 输出一下模型的结构 model.build((None, 70)) model.summary() # 编译模型，配置优化器、损失函数以及监控指标 model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.001), loss=tf.keras.losses.categorical_crossentropy, metrics=['accuracy']) # 调用tf.keras封装的训练接口，开始训练 history = model.fit(x=X_scaled_training, y=Y_scaled_training, epochs=500, validation_data=(X_scaled_testing, Y_scaled_testing)) # 画图 history.history.keys() plt.plot(history.epoch, history.history.get('accuracy'), label='accuracy') plt.plot(history.epoch, history.history.get('val_accuracy'), label='val_accuracy') plt.legend() plt.show()