PrototypicalModel<T, TInput, TOutput>

Models & Types · AiDotNet.MetaLearning.Algorithms

Prototypical model for few-shot classification.

Implements: IModel<TInput, TOutput, ModelMetadata<T>>

For Beginners

After adapting ProtoNets to a new task, you get this model. It classifies new examples instantly by finding the nearest class prototype.

How It Works

The adapted state of ProtoNets for one task: its own copy of the embedding, the class prototypes of the support set, and the learned metric. Predict returns class probabilities, [rows, classes], for Tensor and Matrix outputs, and the most probable class of each example for a Vector output.

Constructors

Constructor Summary
PrototypicalModel(IFullModel<T, TInput, TOutput> featureEncoder, TInput supportInputs, TOutput supportOutputs, ProtoNetsOptions<T, TInput, TOutput> options, INumericOperations<T> numOps) Initializes a new instance of the PrototypicalModel with the paper’s metric.

Properties

Property Summary
ModelMetadata<T> Metadata { get; } Gets the model metadata.

Methods

Method Summary
ModelMetadata<T> GetModelMetadata() Gets metadata about the model.
TOutput Predict(TInput input) Makes predictions using prototype-based classification.
Vector<T> GetParameters() Gets model parameters (not applicable for prototype-based models).
void Train(TInput inputs, TOutput targets) Trains the model (not applicable for prototype-based models).
void UpdateParameters(Vector<T> parameters) Updates model parameters (not applicable for prototype-based models).