Deep.Net


MLP

Namespace: Models

A neural network (multi-layer perceptron) of multiple NeuralLayers and one LossLayer on top.

Nested types and modules

TypeDescription
HyperPars

MLP hyper-parameters.

Pars<'T>

MLP parameters.

Functions and values

Function or valueDescription
loss pars input target
Signature: pars:Pars<'?176151> -> input:ExprT -> target:ExprT -> ExprT
Type parameters: '?176151

Returns an expression for the loss of the neural network with parameters pars given the input input and target values target.

pars mb hp
Signature: mb:ModelBuilder<'T> -> hp:HyperPars -> Pars<'T>
Type parameters: 'T

Creates the parameters for the neural network in the supplied model builder mb using the hyper-parameters hp. See NeuralLayer.pars for documentation about the initialization.

pred pars input
Signature: pars:Pars<'T> -> input:ExprT -> ExprT
Type parameters: 'T

Returns an expression for the output (predictions) of the neural network with parameters pars given the input input.

regualrizationTerm pars input
Signature: pars:Pars<'?176153> -> input:ExprT -> ExprT
Type parameters: '?176153

Calculates sum of all regularization terms of this model.

Fork me on GitHub