ElementwiseLayer

represents a net layer that applies a unary function f to every element of the input tensor.

applies the function specified by "name".

Details and Options

The function f can be any one of the following: Ramp, LogisticSigmoid, Tanh, ArcTan, ArcTanh, Sin, Sinh, ArcSin, ArcSinh, Cos, Cosh, ArcCos, ArcCosh, Log, Exp, Sqrt, Abs, Gamma, LogGamma.
In general, f can be any object that when applied to a single argument gives any combination of Ramp, LogisticSigmoid, etc., together with Plus, Subtract, Times, Divide, Power, Min, Max, Clip, and numbers.
ElementwiseLayer supports the following values for "name":

		"RectifiedLinearUnit" or "ReLU"	Ramp[x]
		"ExponentialLinearUnit" or "ELU"	x when x≥0 Exp[x]-1 when x<0
		"ScaledExponentialLinearUnit" or "SELU"	1.0507 x when x >= 0 1.7581 (Exp[x]-1) when x<0
		"SoftSign"	x/(1+Abs[x])
		"SoftPlus"	Log[Exp[x+1]]
		"HardTanh"	Clip[x,{-1,1}]
		"HardSigmoid"	Clip[(x+1)/2, {0,1}]
		"Sigmoid"	LogisticSigmoid[x]

ElementwiseLayer[…][input] explicitly computes the output from applying the layer.
ElementwiseLayer[…][{input₁,input₂,…}] explicitly computes outputs for each of the input_i.
ElementwiseLayer is typically used inside NetChain, NetGraph, etc.
ElementwiseLayer exposes the following ports for use in NetGraph etc.:
"Input" a numerical tensor of arbitrary rank

"Output" a numerical tensor with the same dimensions as the input
When it cannot be inferred from other layers in a larger net, the option "Input"->{n₁,n₂,…} can be used to fix the input dimensions of ElementwiseLayer.