An Experiment on the Use of Genetic Algorithms for Topology Selection in Deep Learning

Table 1

Search space restrictions.

Description	Value(s)
# convolutional layers	0–3
Convolutional filters	{2, 4, 8, 16, 32}
Kernel size	{3, 5, 7, 9, 11}
Stride	{1, 2, 3, 4}
Pooling	{2, 3, 4}
Activation functions	{Linear, Sigmoid, TanH, ReLU}
# fully connected layers	0–2
Neurons	{8, 16, 32, 64}
Dropout	{0.0, 0.2, 0.4, 0.6}
Activation functions	{Linear, Sigmoid, TanH, ReLU}