DotLayer

✖
`DotLayer`

[Experimental]

✖

DotLayer[]

represents a net layer that takes the dot product of two or more arrays.

✖

DotLayer[{spec₁,spec₂,…}]

uses given transpose specifications for the respective inputs.

Details and Options

DotLayer is typically used inside NetGraph.
DotLayer[…][{in₁,in₂,…}] explicitly computes the output from applying the layer.
The inputs of DotLayer[…] should be arrays, and the output is effectively given by Dot[in₁,in₂,…].
The output of DotLayer[{spec₁,spec₂,…}][{in₁,in₂,…}] is effectively given by Dot[Transpose[in₁,spec₁],Transpose[in₂,spec₂],…].
The transpose specifications spec_i can be:

	None	no transposition
	m<->n	transposes levels m and n
	{m₁<->n₁,m₂<->n₂,…}	sequentially transposes levels m_i and n_i
	{n₁,n₂,…}	transposes so that the k level in the input is the n_k level in the output

Option InputPorts can be used to specify the number, names or shapes of input ports. If InputPorts is not specified, the number of ports and their shapes are inferred from NetChain or NetGraph connectivity.
Besides input ports, DotLayer exposes the following ports for use in NetGraph etc.:
"Output" an array
Within a NetGraph, a DotLayer can be connected using a single edge of the form {src₁,src₂,…}dotlayer, where dotlayer is the name or index of the DotLayer, or as multiple separate edges given in the corresponding order, as src₁dotlayer,src₂dotlayer,…,src_ndotlayer.
The shapes of the inputs to DotLayer (after possible transposition) should be such that the last dimension of input_i is equal to the first dimension of input_i+1.
Options[DotLayer] gives the list of default options to construct the layer. Options[DotLayer[…]] gives the list of default options to evaluate the layer on some data.
Information[DotLayer[…]] gives a report about the layer.
Information[DotLayer[…],prop] gives the value of the property prop of DotLayer[…]. Possible properties are the same as for NetGraph.