def test_relu_2_backward(self):
print('\n==================================')
print(' Test ReLU backward ')
print('==================================')
np.random.seed(123)
relu = ReLU()
x = np.random.randn(7, 7)
d_prev = np.random.randn(*x.shape)
out = relu.forward(x)
dx = relu.backward(d_prev, 0.0)
correct_dx = [[0., -1.29408532, -1.03878821, 0., 0., 0.02968323, 0.],
[0., 1.75488618, 0., 0., 0., 0.79486267, 0.],
[
0., 0., 0.80723653, 0.04549008, -0.23309206,
-1.19830114, 0.19952407
], [0.46843912, 0., 1.16220405, 0., 0., 1.03972709, 0.],
[0., 0., 0., 0., 0., 0., 0.80730819],
[0., -1.0859024, -0.73246199, 0., 2.08711336, 0., 0.],
[
0., 0.18103513, 1.17786194, 0., 1.03111446,
-1.08456791, -1.36347154
]]
e = rel_error(correct_dx, dx)
print('dX relative difference:', e)
self.assertTrue(e <= 5e-08)
def test_relu_1_forward(self):
print('\n==================================')
print(' Test ReLU forward ')
print('==================================')
x = np.linspace(-0.7, 0.5, num=20).reshape(5, 4)
relu = ReLU()
out = relu.forward(x)
correct_out = np.array(
[[0., 0., 0., 0.], [0., 0., 0., 0.], [0., 0., 0., 0.],
[0.05789474, 0.12105263, 0.18421053, 0.24736842],
[0.31052632, 0.37368421, 0.43684211, 0.5]])
e = rel_error(correct_out, out)
print('Relative difference:', e)
self.assertTrue(e <= 5e-08)
def test_model(num_feat, num_classes):
classifier = ClassifierModel()
classifier.add_layer('FC-1', FCLayer(num_feat, 2))
classifier.add_layer('Sigmoid', Sigmoid())
classifier.add_layer('FC-2', FCLayer(2, 5))
classifier.add_layer('ReLU', ReLU())
classifier.add_layer('FC-3', FCLayer(5, 3))
classifier.add_layer('tanh', Tanh())
classifier.add_layer('FC-4', FCLayer(3, num_classes))
classifier.add_layer('Softmax', SoftmaxLayer())
return classifier
def test_model(num_feat, num_classes):
classifier = CNN_Classifier()
classifier.add_layer(
'Conv-1', ConvolutionLayer(num_feat, 2, kernel_size=3, stride=1,
pad=1))
classifier.add_layer('ReLU', ReLU())
classifier.add_layer(
'Conv-2', ConvolutionLayer(2, 3, kernel_size=3, stride=1, pad=1))
classifier.add_layer('tanh', Tanh())
classifier.add_layer(
'Conv-3', ConvolutionLayer(3, 3, kernel_size=3, stride=1, pad=0))
classifier.add_layer('Sigmoid', Sigmoid())
classifier.add_layer('Max-pool - 1',
MaxPoolingLayer(kernel_size=2, stride=1))
classifier.add_layer('FC-4', FCLayer(12, num_classes))
classifier.add_layer('Softmax', SoftmaxLayer())
return classifier
# CNN 모델을 구성하고 실험하세요 #
# 레이어는 아래 예제와 같이 추가할 수 있습니다. #
# 주의하세요! 레이어들은 추가한 순서대로 실행됩니다. #
# Layer에 따라, 구현에 따라 다르지만 1 Epoch에 5분 이상 걸려도 정상입니다. #
# ^^ #
# #
###########################################################################
# Add Layers, Layer 추가 예제
# 아래 예시를 참고하여 과제에 주어진 대로 (혹은 과제와 별개로 원하는 대로) Layer를 추가하세요.
# Convolution Layer
CNN.add_layer('Conv Layer - 1', ConvolutionLayer(in_channels=in_channnel, out_channels=8, kernel_size=3, pad=1))
# ReLU Layer
CNN.add_layer('ReLU Layer - 1', ReLU())
# Convolution Layer
CNN.add_layer('Conv Layer - 2', ConvolutionLayer(in_channels=8, out_channels=8, kernel_size=3, pad=1))
# ReLU Layer
CNN.add_layer('ReLU Layer - 2', ReLU())
# Max-pool Layer
CNN.add_layer('Max-Pool Layer', MaxPoolingLayer(kernel_size=2, stride=2))
# FC Layer
CNN.add_layer('FC Example Layer - 1', FCLayer(input_dim=1568, output_dim=500))
# FC Layer
CNN.add_layer('FC Example Layer - 2', FCLayer(input_dim=500, output_dim=5))
#plt.show()
# ===========================================================================
print(
'================================ 2. ReLU =================================='
)
"""
* Correct ReLU *
Forward:
[1. 0. 3.]
Backward:
[-10 0 -30]
"""
relu = ReLU()
temp2 = np.array([1, -0.1, 3], dtype=np.float32)
temp3 = np.array([-10, -20, -30], dtype=np.float32)
print('ReLU Check')
print('Forward: \n', relu.forward(temp2))
print('Backward: \n', relu.backward(temp3))
print()
# ===========================================================================
print(
'=========================== 3. Convolution Layer =========================='
)
# Convolution with stride 1, no padding
in_channel = 1
out_channel = 3
ex) shuffle data every epoch (You don't need to care this. But you can try personally.)
ex) set/change random seed (I've already done it.)
"""
print('===== ReLU Check =====')
"""
The results should be exactly same as below:
결과는 아래와 일치해야 합니다:
Forward:
[1. 0. 3.]
Backward:
[-10 0 -30]
"""
relu = ReLU()
temp2 = np.array([1, -0.1, 3])
temp3 = np.array([-10, -20, -30])
print('Forward: \n', relu.forward(temp2))
print('Backward: \n', relu.backward(temp3))
print()
print('===== Sigmoid Check =====')
"""
The results should be exactly same as below:
결과는 아래와 일치해야 합니다:
Forward:
[[0.26894142 0.88079708 0.62245933]
[0.5 0.47502081 0.52497919]]
# Enjoy. #
# #
# CNN 모델을 구성하고 실험하세요 #
# 레이어는 아래 예제와 같이 추가할 수 있습니다. #
# 주의하세요! 레이어들은 추가한 순서대로 실행됩니다. #
# Layer에 따라, 구현에 따라 다르지만 1 Epoch에 5분 이상 걸려도 정상입니다. #
# ^^ #
# #
###########################################################################
# Add Layers, Layer 추가 예제
# 아래 예시를 참고하여 과제에 주어진 대로 (혹은 과제와 별개로 원하는 대로) Layer를 추가하세요.
# Convolution Layer
CNN.add_layer('Conv Layer 1', ConvolutionLayer(in_channels=in_channnel, out_channels=8, kernel_size=3, pad=1))
CNN.add_layer('ReLU 1', ReLU())
CNN.add_layer('Conv Layer 2', ConvolutionLayer(in_channels=8, out_channels=8, kernel_size=3, pad=1))
CNN.add_layer('ReLU 2', ReLU())
# Max-pool Layer
CNN.add_layer('Max-Pool Layer', MaxPoolingLayer(kernel_size=2, stride=2))
# FC Layer
CNN.add_layer('FC Layer 1', FCLayer(input_dim=1568, output_dim=500))
CNN.add_layer('FC Layer 2', FCLayer(input_dim=500, output_dim=5))
# Softmax Layer
# 이 layer는 항상 마지막에 추가
CNN.add_layer('Softmax Layer', SoftmaxLayer())
# Model Architecture 출력
# #
###########################################################################
# Add Layers, Layer 추가 예제
# 아래 예시를 참고하여 과제에 주어진 대로 (혹은 과제와 별개로 원하는 대로) Layer를 추가하세요.
# Convolution Layer
CNN.add_layer(
'Conv Example Layer',
ConvolutionLayer(in_channels=in_channnel,
out_channels=8,
kernel_size=3,
pad=1))
# ReLU Layer
CNN.add_layer('ReLU Example Layer', ReLU())
# Convolution Layer
CNN.add_layer(
'Conv Example Layer2',
ConvolutionLayer(in_channels=8, out_channels=8, kernel_size=3, pad=1))
# ReLU Layer
CNN.add_layer('ReLU Example Layer2', ReLU())
# Max-pool Layer
CNN.add_layer('Max-Pool Example Layer', MaxPoolingLayer(kernel_size=2,
stride=2))
# FC Layer
CNN.add_layer('FC Example Layer1', FCLayer(input_dim=1568, output_dim=500))
num_train, in_channels, H, W = x_train.shape
num_class = y_train.shape[1]
train_accuracy = []
valid_accuracy = []
best_epoch = -1
best_acc = -1
best_model = None
# =============================== EDIT HERE ===============================
# Add layers
CNN.add_layer('Conv-1', ConvolutionLayer(in_channels=in_channels, out_channels=4, kernel_size=3, pad=1))
CNN.add_layer('ReLU-1',ReLU())
CNN.add_layer('Conv-2', ConvolutionLayer(in_channels=4, out_channels=4, kernel_size=3, pad=1))
CNN.add_layer('ReLU-2',ReLU())
CNN.add_layer('Max-pool-1',MaxPoolingLayer(2,2))
CNN.add_layer('FC-1',FCLayer(784,500))
CNN.add_layer('ReLU-3',ReLU())
CNN.add_layer('FC-2',FCLayer(500,10))
CNN.add_layer('Softmax Layer',SoftmaxLayer())
# =========================================================================
CNN.summary()
print('Training Starts...')
num_batch = int(np.ceil(num_train / batch_size))