MLPNTMController<T, TInput, TOutput>

Models & Types · AiDotNet.MetaLearning.Algorithms

MLP-based NTM controller implementation with learnable parameters.

Implements: INTMController<T>

How It Works

This controller uses a multi-layer perceptron to process inputs and generate addressing parameters for the NTM memory. Unlike LSTM, MLP is stateless and processes each timestep independently.

Architecture:

Constructors

Constructor Summary
MLPNTMController(NTMOptions<T, TInput, TOutput> options) Initializes a new instance of MLPNTMController with learnable weights.

Methods

Method Summary
IReadOnlyList<Tensor<T>> GetParameterTensors()
List<NTMAddressingParameters<T>> GenerateReadAddressing(Tensor<T> output)
List<Tensor<T>> GenerateReadKeys(Tensor<T> output)
MLPNTMController<T, TInput, TOutput> Clone() Creates an independent deep copy of this controller, preserving its resolved input width, learned parameters, and cached hidden state.
NTMAddressingParameters<T> GenerateWriteAddressing(Tensor<T> output)
Tensor<T> Forward(Tensor<T> input, List<Tensor<T>> readContents)
Tensor<T> GenerateAddVector(Tensor<T> output)
Tensor<T> GenerateEraseVector(Tensor<T> output)
Tensor<T> GenerateOutput(Tensor<T> output, List<Tensor<T>> readContents)
Tensor<T> GenerateWriteKey(Tensor<T> output)
Vector<T> GetParameters()
void Reset()
void SetParameters(Vector<T> parameters)