# ------[Data Acquisition n Preprocessing]------
dataset, label = noise_time_shift_xcor_return(time_axis, fs=fs, num_series=100)
train_x, train_y, test_x, test_y = break_balanced_class_into_train_test(
input=dataset,
label=label,
num_classes=num_classes,
train_split=0.7,
verbose=True)
# reshape to satisfy conv2d input shape
train_x, train_y, test_x, test_y = reshape_3d_to_4d_tocategorical(
train_x,
train_y,
test_x,
test_y,
fourth_dim=1,
num_classes=num_classes,
verbose=True)
model = cnn_51_159_3class_v1()
model.compile(optimizer='adam',
loss='categorical_crossentropy',
metrics=['accuracy'])
model_logger = ModelLogger(model, model_name='cnn_51_159')
history = model.fit(x=train_x,
y=train_y,
batch_size=30,
epochs=100,
verbose=1,
validation_data=(test_x, test_y))
plt.subplot(221)
plt.imshow(X_train[10], cmap=plt.get_cmap('gray'))
plt.subplot(222)
plt.imshow(X_train[11], cmap=plt.get_cmap('gray'))
plt.subplot(223)
plt.imshow(X_train[12], cmap=plt.get_cmap('gray'))
plt.subplot(224)
plt.imshow(X_train[13], cmap=plt.get_cmap('gray'))
# show the plot
plt.show()
X_train = X_train / 255
X_test = X_test / 255
train_x, train_y, test_x, test_y = reshape_3d_to_4d_tocategorical(
X_train, y_train, X_test, y_test, num_classes=10, verbose=True)
model = cnn_28_28_mnist_10class()
model.compile(optimizer='adam',
loss='categorical_crossentropy',
metrics=['accuracy'])
# logging
model_logger = ModelLogger(model, model_name='CNN_MNIST_28_28')
# tensorboard
# tb_callback = TensorBoard(log_dir='./Graph', histogram_freq=0, write_graph=True, write_images=True)
time_start = time.time()
# train
history = model.fit(x=train_x,
y=train_y,