In convolutional() in feedforward_nn.py, sigmoid activation should be used for binary classification

Palashio / libra

Ergonomic machine learning for everyone.

http://libradocs.org/

MIT License

1.91k stars 109 forks source link

In convolutional() in feedforward_nn.py, sigmoid activation should be used for binary classification #327

Closed abx393 closed 4 years ago

abx393 commented 4 years ago

By default, the output layer has num_classes nodes and softmax activation is being used on the final layer for CNN models in convolutional(). If num_classes is 2, sigmoid activation should be used and the number of nodes in the final layer should be 1, not 2.