>>> lenet <- mx.symbol.Variable("data") %>%
# 첫 번째 합성곱층(Convolutional Layer Set 1 (Conv > Tanh > Pool) )
>>>      mx.symbol.Convolution(kernel=c(5, 5), num_filter=20, name="Conv1") %=>% "Conv1" %>%
>>>      mx.symbol.Activation(act_type="tanh", name="Act1") %=>% "Act1" %>%
>>>      mx.symbol.Pooling( pool_type="max", kernel=c(2, 2), stride=c(2,2), name = "Pool1") %=>% "Pool1" %>%

# 두 번째 합성곱층(Convolutional Layer Set 1(Conv > Tanh > Pool))
>>>      mx.symbol.Convolution(kernel=c(5, 5), num_filter=50, name="Conv2") %=>% "Conv2" %>%
>>>      mx.symbol.Activation(act_type="tanh", name="Act2") %=>% "Act2" %>%
>>>      mx.symbol.Pooling(pool_type="max", kernel=c(2, 2), stride=c(2, 2), name = "Pool2") %=>% "Pool2" %>%
# 2차원 필터를 1차원으로 변경시키기
>>>      mx.symbol.flatten(name="Flat") %=>% "Flat1" %>%

# 함수 FulltConnected로 마지막 은닉층과 연결
>>>      mx.symbol.FullyConnected(num_hidden=500, name="Full1") %=>% "Full1" %>%
>>>      mx.symbol.Activation(act_type="tanh", name="Act3") %=>% "Act3" %>%

# 함수 FulltConnected로 마지막 결과층과 연결
>>>      mx.symbol.FullyConnected(num_hidden=10, name="Full2") %=>% "Full2" %>%
>>>      mx.symbol.SoftmaxOutput(name="SoftM") %=>% "SoftM"

>>> graph.viz(lenet, direction = "LR")