Layer Types

This page specifies the different layer types that can be used and their function signatures.

;;Default layer
:dense (fn [] ...)

;;Default output layer
:output (fn [loss-fn] ...)

;;Restricted Boltzmann Machine layer
:rbm (fn [] ...)

;;Bidirectional RNN layer
:graves-bidirectional-lstm (fn [] ...)

;;Default RNN layer
:graves-lstm (fn [] ...)

;;Specific output layer used by Recurrent Neural Nets
:rnn-output (fn [loss-fn] ...)

;;Convolutional layer with optional parameters
:convolution (fn 
               ([kernel-size] ...)
               ([kernel stride] ...)
               ([kernel stride pad] ...))

;;Sub sampling layer
:sub-sampling (fn 
                ([pooling-type] ... )
                ([kernel-size stride] ... ))

;;Local response normalization layer
:local-response-normalization (fn [] ...)