ETPNAlgorithm<T, TInput, TOutput>
Models & Types · AiDotNet.MetaLearning.Algorithms
Implementation of ETPN: Embedding-Transformed Prototypical Networks.
Inheritance: ModelBase<T, TInput, TOutput> → MetaLearnerBase<T, TInput, TOutput> → ETPNAlgorithm<T, TInput, TOutput>
Implements: IGeneratedParameterRegistrar<T>
How It Works
ETPN learns a task-specific embedding transformation applied transductively. The transformation is computed from both support and query gradient information, enabling the adapted parameter space to be more discriminative for the specific task. A learned projection maps combined gradient features to per-parameter scaling factors.
Algorithm:
Constructors
| Constructor | Summary |
|---|---|
ETPNAlgorithm(ETPNOptions<T, TInput, TOutput> options) |
Properties
| Property | Summary |
|---|---|
MetaLearningAlgorithmType AlgorithmType { get; } |
Methods
| Method | Summary |
|---|---|
IModel<TInput, TOutput, ModelMetadata<T>> Adapt(IMetaLearningTask<T, TInput, TOutput> task) |
|
T MetaTrain(TaskBatch<T, TInput, TOutput> taskBatch) |