cnn_classify.py

import uwb_dataset
import numpy as np
from keras.models import Sequential
from keras.layers import Dense, Embedding, Dropout, Conv1D, MaxPooling1D, GlobalMaxPooling1D
from keras.callbacks import EarlyStopping

# import raw data
data = uwb_dataset.import_from_files()

# divide CIR by RX preable count (get CIR of single preamble pulse)
# item[2] represents number of acquired preamble symbols
for item in data:
    item[15:] = item[15:] / float(item[2])

# test data
train = data[:30000, :]
np.random.shuffle(train)
x_train = train[:30000, 15:]
y_train = train[:30000, 0]
x_test = data[30000:, 15:]
y_test = data[30000:, 0]

# feed data
x_val = x_train[25000:]
y_val = y_train[25000:]
x_train = x_train[:25000]
y_train = y_train[:25000]

# x_train = np.reshape(x_train, (x_train.shape[0], x_train.shape[1], 1)).transpose(2,0,1)
# x_val = np.reshape(x_val, (x_val.shape[0], x_val.shape[1], 1)).transpose(2,0,1)
# x_test = np.reshape(x_test, (x_test.shape[0], x_test.shape[1], 1)).transpose(2,0,1)

# total 1016 data for CIR
# define model
# define model 模型定义
model = Sequential()
model.add(Embedding(2700, 1024, input_length=1016))
model.add(Conv1D(filters=10, kernel_size=4, padding='valid', activation='relu'))
model.add(Conv1D(20, 5, padding='valid', activation='relu'))
model.add(MaxPooling1D(pool_size=2, strides=2, padding='valid'))
model.add(Conv1D(20, 4, padding='valid', activation='relu'))
model.add(Conv1D(40, 4, padding='valid', activation='relu'))
model.add(GlobalMaxPooling1D())
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(1, activation='sigmoid'))
print(model.summary())

# compile 编译与训练
# loss='binary_crossentropy'：二分类交叉熵损失函数，用于二分类问题。
# optimizer='adam'：使用Adam优化器进行梯度更新。
# metrics=['accuracy']：模型训练过程中会跟踪准确率。
model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])

#早期停止
early_stopping = EarlyStopping(monitor='val_loss', patience=3, restore_best_weights=True)
hist = model.fit(x_train, y_train, epochs=15, batch_size=64, validation_data=(x_val, y_val), callbacks=[early_stopping])


# evaluation 评估
loss_and_metrics = model.evaluate(x_test, y_test, batch_size=64)
print('## evaluation loss and metrics ##')
print(loss_and_metrics)

# save
model.save('CNN_Classify.h5')