Skip to main content

Table 2 Model hyperparameters

From: Convolutional neural network based on SMILES representation of compounds for detecting chemical motif

Hyperparameter Considered values
1st convolution
 No. of filters [1,1024]
 Window size [1,51]
 Stride size {1,3,5}
 Padding {None, Half of window size}
1st pooling
 Type {Max, Average}
 Window size [1,51]
 Stride size {1,3,5}
 Padding {None, Half of window size}
2nd convolution
 No. of filters [1,1024]
 Window size [1,51]
 Stride size {1,3,5}
 Padding {None, Half of window size}
2nd pooling
 Type {Max, Average}
 Window size [1,51]
 Stride size {1,3,5}
 Padding {None, Half of window size}
Global pooling {None, Max pooling}
Output layer {softmax, sigmoid}
Activation function {ReLU, Leaky ReLU, Parametric ReLU}
Minibatch size {32, 64, 128, 256, 512}
Batch normalization {None, after conv.}
Dropout {None, before output}
Optimizer {Adam, AdaGrad}
Learning rate {0.0001, 0.001, 0.01, 0.1}
Loss function {Mean squared error, Cross entropy}