InputPorts

is an option to specify the number, names or shapes of input ports for some neural net layers.

Details

InputPorts is typically used in layers that support an arbitrary number of input ports, such as ThreadingLayer, CatenateLayer, DotLayer and PlaceholderLayer.
Typical settings for InputPorts include:

	Automatic	infer number of ports and their shapes from NetChain or NetGraph connectivity
	n	n ports with default names and unspecified shapes
	{"name₁","name₂",…}	explicit port names with unspecified shapes
	{shape₁,shape₂,…}	default port names with specified shapes or encoders
	<\|"name₁"shape₁,…\|>	explicit port names with specified shapes or encoders

	"Real"	a single real number
	"Integer"	a single integer
	Restricted["Integer",n]	an integer between 1 and n
	n	a vector of length n
	{n₁,n₂,…}	an array of dimensions n₁×n₂×…
	"Varying"	a vector whose length is variable
	{"Varying",n₂,n₃,…}	an array whose first dimension is variable and remaining dimensions are n₂×n₃×…
	NetEncoder[…]	an encoder

open allclose all

By default, multi-input layers such as ThreadingLayer have an undefined number of input ports:

Use InputPorts to create a ThreadingLayer with three ports having default names and unspecified shapes:

Create a CatenateLayer with two inputs having explicit names and unspecified shapes:

Create a PlaceholderLayer with four inputs having default names and specified shapes:

Create a DotLayer with two inputs having explicit names and specified shapes:

Top