机器学习-信号处理--基于DEAP数据集的情绪分类的典型深度学习模型构建

关于

本实验采用DEAP情绪数据集进行数据分类任务。使用了三种典型的深度学习网络：2D 卷积神经网络；1D卷积神经网络+GRU； LSTM网络。

工具

数据集

DEAP数据

图片来源： DEAP: A Dataset for Emotion Analysis using Physiological and Audiovisual Signals

方法实现

2D-CNN网络

加载必要库函数

import pandas as pd
import keras.backend as K
import numpy as np
import pandas as pd
from keras.models import Sequential
from keras.layers import Dense
from keras.models import Sequential
from keras.layers.convolutional import Conv1D
from keras.layers.convolutional import MaxPooling1D
from tensorflow.keras.utils import to_categorical 
from keras.layers import Flatten
from keras.layers import Dense
import numpy as np
import keras
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPooling2D
from keras import backend as K
from keras.models import Model
import timeit
from keras.models import Sequential
from keras.layers.core import Flatten, Dense, Dropout
from keras.layers.convolutional import Convolution1D, MaxPooling1D, ZeroPadding1D
from tensorflow.keras.optimizers import SGD
#import cv2, numpy as np
import warnings
warnings.filterwarnings('ignore')

加载DEAP数据集

data_training = []
label_training = []
data_testing = []
label_testing = []

for subjects in subjectList:

    with open('/content/drive/My Drive/leading_ai/try/s' + subjects + '.npy', 'rb') as file:
      sub = np.load(file,allow_pickle=True)
      for i in range (0,sub.shape[0]):
        if i % 5 == 0:
          data_testing.append(sub[i][0])
          label_testing.append(sub[i][1])
        else:
          data_training.append(sub[i][0])
          label_training.append(sub[i][1])

np.save('/content/drive/My Drive/leading_ai/data_training', np.array(data_training), allow_pickle=True, fix_imports=True)
np.save('/content/drive/My Drive/leading_ai/label_training', np.array(label_training), allow_pickle=True, fix_imports=True)
print("training dataset:", np.array(data_training).shape, np.array(label_training).shape)

np.save('/content/drive/My Drive/leading_ai/data_testing', np.array(data_testing), allow_pickle=True, fix_imports=True)
np.save('/content/drive/My Drive/leading_ai/label_testing', np.array(label_testing), allow_pickle=True, fix_imports=True)
print("testing dataset:", np.array(data_testing).shape, np.array(label_testing).shape)

 数据标准化

from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
x_train = scaler.fit_transform(x_train)
x_test = scaler.fit_transform(x_test)

定义训练超参数

batch_size = 256
num_classes = 10
epochs = 200
input_shape=(x_train.shape[1], 1)

 定义模型

from keras.layers import Convolution1D, ZeroPadding1D, MaxPooling1D, BatchNormalization, Activation, Dropout, Flatten, Dense
from keras.regularizers import l2

model = Sequential()
intput_shape=(x_train.shape[1], 1)
model.add(Conv1D(164, kernel_size=3,padding = 'same',activation='relu', input_shape=input_shape))
model.add(BatchNormalization())
model.add(MaxPooling1D(pool_size=(2)))
model.add(Conv1D(164,kernel_size=3,padding = 'same', activation='relu'))
model.add(BatchNormalization())
model.add(MaxPooling1D(pool_size=(2)))
model.add(Conv1D(82,kernel_size=3,padding = 'same', activation='relu'))
model.add(MaxPooling1D(pool_size=(2)))
model.add(Flatten())
model.add(Dense(82, activation='tanh'))
model.add(Dropout(0.2))
model.add(Dense(42, activation='tanh'))
model.add(Dropout(0.2))
model.add(Dense(21, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(num_classes, activation='softmax'))
model.summary()

模型配置和训练

model.compile(loss=keras.losses.categorical_crossentropy,
              optimizer='adam',
              metrics=['accuracy'])



history=model.fit(x_train, y_train,
          batch_size=batch_size,
          epochs=epochs,  
          verbose=1,validation_data=(x_test,y_test))

模型测试集验证

score = model.evaluate(x_test, y_test, verbose=1)
print('Test loss:', score[0])
print('Test accuracy:', score[1])

模型训练过程可视化

# summarize history for accuracy
plt.plot(history.history['accuracy'])
plt.plot(history.history['val_accuracy'])
plt.title('model accuracy')
plt.ylabel('accuracy')
plt.xlabel('epoch')
plt.legend(['train', 'test'], loc='upper left')
plt.show()

模型测试集分类混沌矩阵

cmatrix=confusion_matrix(y_test1, y_pred)

import seaborn as sns
figure = plt.figure(figsize=(8, 8))
sns.heatmap(cmatrix, annot=True,cmap=plt.cm.Blues)
plt.tight_layout()
plt.ylabel('True label')
plt.xlabel('Predicted label')
plt.show()

模型测试集分类report

from sklearn import metrics
y_pred = np.around(model.predict(x_test))
print(metrics.classification_report(y_test,y_pred))

1D-CNN+GRU网络

数据预处理

必要库函数加载,数据加载预处理，同2D CNN一样，不在赘述。

!pip install git+https://github.com/forrestbao/pyeeg.git
import numpy as np
import pyeeg as pe
import pickle as pickle
import pandas as pd
import matplotlib.pyplot as plt
import math

import os
import time
import timeit
import keras
import keras.backend as K
from keras.models import Model
from keras.layers import Flatten
from keras.datasets import mnist
from keras.models import Sequential
from sklearn.preprocessing import normalize
from tensorflow.keras.optimizers import SGD
from keras.layers.convolutional import Conv1D
from keras.layers.convolutional import MaxPooling1D
from keras.layers.convolutional import ZeroPadding1D
from tensorflow.keras.utils import to_categorical
from keras.layers import Dense, Dropout, Flatten,GRU

import warnings
warnings.filterwarnings('ignore')

模型搭建

from keras.layers import Convolution1D, ZeroPadding1D, MaxPooling1D, BatchNormalization, Activation, Dropout, Flatten, Dense,GRU,LSTM
from keras.regularizers import l2

from keras.models import load_model
from keras.layers import Lambda
import tensorflow as tf



model_2 = Sequential()

model_2.add(Conv1D(128, 3, activation='relu', input_shape=input_shape))
model_2.add(MaxPooling1D(pool_size=2))
model_2.add(Dropout(0.2))

model_2.add(Conv1D(128, 3,  activation='relu'))
model_2.add(MaxPooling1D(pool_size=2))
model_2.add(Dropout(0.2))

model_2.add(GRU(units = 256, return_sequences=True))  
model_2.add(Dropout(0.2))

model_2.add(GRU(units = 32))
model_2.add(Dropout(0.2))

model_2.add(Flatten())

model_2.add(Dense(units = 128, activation='relu'))
model_2.add(Dropout(0.2))

model_2.add(Dense(units = num_classes))
model_2.add(Activation('softmax'))

model_2.summary()

模型编译和训练

model_2.compile(
    optimizer ="adam",
    loss = 'categorical_crossentropy',
    metrics=["accuracy"]
)



history_2 = model_2.fit(
    x_train, 
    y_train,
    epochs=epochs,
    batch_size=batch_size,
    verbose=1,
    validation_data=(x_test, y_test),
    callbacks=[
        keras.callbacks.EarlyStopping(
            monitor='val_loss',
            patience=20,
            restore_best_weights=True
        )
    ]
)

 模型训练过程可视化

# summarize history for accuracy
plt.plot(history_2.history['accuracy'],color='green',linewidth=3.0)
plt.plot(history_2.history['val_accuracy'],color='red',linewidth=3.0)
plt.title('model accuracy')
plt.ylabel('accuracy')
plt.xlabel('epoch')
plt.legend(['train', 'test'], loc='upper left')

plt.savefig("/content/drive/My Drive/GRU/model accuracy.png")
plt.show()


# summarize history for loss
plt.plot(history_2.history['loss'],color='green',linewidth=2.0)
plt.plot(history_2.history['val_loss'],color='red',linewidth=2.0)
plt.title('model loss')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train', 'test'], loc='upper left')

plt.savefig("/content/drive/My Drive/GRU/model loss.png")
plt.show()

 模型测试集分类混沌矩阵和分类report

LSTM网络

数据加载/预处理

同上

模型搭建和训练

  from keras.regularizers import l2
  from keras.layers import Bidirectional
  from keras.layers import LSTM

  model = Sequential()
 
  model.add(Bidirectional(LSTM(164, return_sequences=True), input_shape=input_shape))
  model.add(Dropout(0.6))

  model.add(LSTM(units = 256, return_sequences = True))  
  model.add(Dropout(0.6))

  model.add(LSTM(units = 82, return_sequences = True))  
  model.add(Dropout(0.6))

  model.add(LSTM(units = 82, return_sequences = True))  
  model.add(Dropout(0.4))

  model.add(LSTM(units = 42))
  model.add(Dropout(0.4))

  model.add(Dense(units = 21))
  model.add(Activation('relu'))

  model.add(Dense(units = num_classes))
  model.add(Activation('softmax'))
  model.compile(optimizer ="adam", loss =keras.losses.categorical_crossentropy,metrics=["accuracy"])
  model.summary()
  m=model.fit(x_train, y_train,epochs=200,batch_size=256,verbose=1,validation_data=(x_test, y_test))

模型训练过程可视化

import matplotlib.pyplot as plt
print(m.history.keys())
# summarize history for accuracy
plt.plot(m.history['accuracy'],color='green',linewidth=3.0)
plt.plot(m.history['val_accuracy'],color='red',linewidth=3.0)

plt.title('model accuracy')
plt.ylabel('accuracy')
plt.xlabel('epoch')
plt.legend(['train', 'test'], loc='upper left')

plt.savefig("./Bi- LSTM/model accuracy.png")
plt.show()


import imageio
plt.plot(m.history['loss'],color='green',linewidth=2.0)
plt.plot(m.history['val_loss'],color='red',linewidth=2.0)

plt.title('model loss')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train', 'test'], loc='upper left')

#to save the image
plt.savefig("./Bi- LSTM/model loss.png")
plt.show()

模型测试集分类性能

代码获取

后台私信，请注明文章题目（数据需要自己下载和处理）

相关项目和代码问题，欢迎交流。