search

snuspl / dolphin

14 stars 2 forks source link

Convolutional layer base implementation #170

Closed ghost closed 8 years ago

ghost commented 8 years ago

This PR introduce new layer class ConvolutionalLayer. This layer extends LayerBase, and include implementation for isLearnable(), getOutputShape() and generateParameterGradient(). This PR include some modifications in LayerParameter, which enable ConvolutionalLayer to make use of its new parameters, paddingHeight and paddingWidth. Also, ConvolutionalLayerConfigurationBuilder class and ConvolutionalLayerParameterInitializer class are added for configuration of convolutional layer. neural_network.proto is modified to take inputs for convolutional layer. Concrete implementation for feedForward() and backPropagate() will be made in later PRs.

beomyeol commented 8 years ago

@seyounglee95 I've done my pass. Please check comments that I left. :)

ghost commented 8 years ago

@beomyeol Thank you for your comments. I pushed a commit. Please take a look.

beomyeol commented 8 years ago

@seyounglee95 I've done my pass and left some comments. I'll merge this after they are resolved.

ghost commented 8 years ago

@beomyeol Thank you for your comments. Please take a look.

beomyeol commented 8 years ago

@seyounglee95 This looks good. I'm merging.