MPTSAlgorithm<T, TInput, TOutput>
Models & Types · AiDotNet.MetaLearning.Algorithms
Implementation of MPTS: Meta-learning with Progressive Task-Specific adaptation.
Inheritance: ModelBase<T, TInput, TOutput> → MetaLearnerBase<T, TInput, TOutput> → MPTSAlgorithm<T, TInput, TOutput>
Implements: IGeneratedParameterRegistrar<T>
How It Works
MPTS divides model parameters into groups and progressively unfreezes them during the inner loop. High-priority groups (typically the classifier head) are adapted from the first step, while lower-priority groups (backbone) are gradually unfrozen as adaptation progresses. Learned priority scores determine the unfreezing schedule.
Algorithm:
Constructors
| Constructor | Summary |
|---|---|
MPTSAlgorithm(MPTSOptions<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) |