Class ComputationGraph<T>

java.lang.Object

org.episteme.core.mathematics.ml.neural.ComputationGraph<T>

Type Parameters:

T - the data type.

public class ComputationGraph<T>
extends Object

Represents a neural network as a sequence of layers (Computational Graph).

This class orchestrates the forward and backward passes across all layers.

Since:

2.0

Author:

Silvere Martin-Michiellot, Gemini AI (Google DeepMind)

Constructor Summary

Constructors

Constructor	Description
ComputationGraph()

Method Summary

Modifier and Type	Method	Description
ComputationGraph<T>	add(Layer<T> layer)	Adds a layer to the network.
List<Layer<T>>	getLayers()	Returns the list of layers.
Tensor<T>	predict(Tensor<T> input)	Performs inference (forward pass) on the input.
void	trainStep(Tensor<T> input, Tensor<T> target, Optimizer<T> optimizer)	Performs a training step (forward + backward + optimize).

Methods inherited from class Object

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

Constructor Details

ComputationGraph

public ComputationGraph()

Method Details

add

public ComputationGraph<T> add(Layer<T> layer)

Adds a layer to the network.

Parameters:

layer - the layer to add.

Returns:

this graph for chaining.

predict

public Tensor<T> predict(Tensor<T> input)

Performs inference (forward pass) on the input.

Parameters:

input - the input tensor.

Returns:

the output tensor (prediction).

trainStep

public void trainStep(Tensor<T> input,
 Tensor<T> target,
 Optimizer<T> optimizer)

Performs a training step (forward + backward + optimize).

Parameters:

input - input batch.

target - target values.

optimizer - optimizer to update weights.

getLayers

public List<Layer<T>> getLayers()

Returns the list of layers.