Package io.github.kirstenali.deepj.layers

Class FNN

java.lang.Object

io.github.kirstenali.deepj.layers.FNN

All Implemented Interfaces:

Layer, Trainable

public final class FNN
extends Object
implements Layer

Flexible fully-connected neural network (MLP) built from Linear projections.

This class is intentionally small: it exists as a convenience for users who want a classic ANN-style model while deepj remains transformer-oriented.

Design notes:

• Uses Linear so parameters can be optimized externally (AdamW/SGD/etc.).
• Accepts an ActivationFunction factory to avoid state-sharing bugs during backprop.

Constructor Summary

Constructors

Constructor	Description
FNN(int inputSize, int[] hiddenSizes, int outputSize, Supplier<ActivationFunction> hiddenActivationFactory, ActivationFunction outputActivation, Random rnd)	Build an MLP of the form: Linear -> act -> Linear -> act -> ...
FNN(int inputSize, int[] hiddenSizes, int outputSize, Supplier<ActivationFunction> hiddenActivationFactory, Random rnd)

Method Summary

Modifier and Type	Method	Description
Tensor	backward(Tensor gradOut)
Tensor	forward(Tensor x)
List<Parameter>	parameters()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface io.github.kirstenali.deepj.training.Trainable

zeroGrad

Constructor Details

FNN

public FNN(int inputSize,
 int[] hiddenSizes,
 int outputSize,
 Supplier<ActivationFunction> hiddenActivationFactory,
 ActivationFunction outputActivation,
 Random rnd)

Build an MLP of the form: Linear -> act -> Linear -> act -> ... -> Linear -> (optional outputAct)

FNN

public FNN(int inputSize,
 int[] hiddenSizes,
 int outputSize,
 Supplier<ActivationFunction> hiddenActivationFactory,
 Random rnd)

Method Details

forward

public Tensor forward(Tensor x)

Specified by:

forward in interface Layer

backward

public Tensor backward(Tensor gradOut)

Specified by:

backward in interface Layer

parameters

public List<Parameter> parameters()

Specified by:

parameters in interface Trainable