CNN - LaNet-5 model

[iammiori]

CNN - LaNet-5 model

[iammiori]

layer 쌓는 부분만 기재.
code#1

# Layer 1 :  Convolution
# Input = 32x32x1. 
# Output = 28x28x6
# w_filter_size = x
# 32 - x  + 1 = 28 
# finally, x = 5

conv1_w = tf.Variable(tf.random_normal(shape = [5,5,1,6], stddev = 0.01))
conv1 = tf.nn.conv2d(X,conv1_w, strides = [1,1,1,1], padding = 'VALID')
conv1 = tf.nn.relu(conv1)

# Subsampling means pooling. 
# Input = 28x28x6. 
# Output = 14x14x6.

pool_1 = tf.nn.max_pool(conv1,ksize = [1,2,2,1], strides = [1,2,2,1], padding = 'VALID')

#---------------------------------------------------

# Layer 2 :  Convolution
# Input = 14x14x6. 
# Output = 10x10x16

conv2_w = tf.Variable(tf.random_normal(shape = [5,5,6,16], stddev = 0.01))
conv2 = tf.nn.conv2d(pool_1, conv2_w, strides = [1,1,1,1], padding = 'VALID') 
conv2 = tf.nn.relu(conv2)
 
# Subsampling means pooling. 
# Input = 10x10x16. 
# Output = 5x5x16.= 400

pool_2 = tf.nn.max_pool(conv2, ksize = [1,2,2,1], strides = [1,2,2,1], padding = 'VALID') 
fc1 = flatten(pool_2)

#-------------------------------------------------------------

# Layer 3: Fully Connected. 
# Input = 400. 
# Output = 120.
fc1_w = tf.Variable(tf.random_normal(shape = (400,120), stddev = 0.01))
fc1 = tf.matmul(fc1,fc1_w) 
fc1 = tf.nn.relu(fc1)

# Layer 4: Fully Connected. 
# Input = 120. 
# Output = 84.
fc2_w = tf.Variable(tf.random_normal(shape = (120,84), stddev = 0.01))
fc2 = tf.matmul(fc1,fc2_w) 
fc2 = tf.nn.relu(fc2)

# Layer 5: Fully Connected. 
# Input = 84. Output = 10.
fc3_w = tf.Variable(tf.random_normal(shape = (84,10),  stddev = 0.01))
logits = tf.matmul(fc2, fc3_w) 

    
# define cost/loss & optimizer
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(
    logits=logits, labels=Y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)