DiscoRLAlgorithm<T, TInput, TOutput>
Models & Types · AiDotNet.MetaLearning.Algorithms
Implementation of DiscoRL: Discovery-based meta-RL with reusable skill discovery.
Inheritance: ModelBase<T, TInput, TOutput> → MetaLearnerBase<T, TInput, TOutput> → DiscoRLAlgorithm<T, TInput, TOutput>
Implements: IGeneratedParameterRegistrar<T>
How It Works
DiscoRL discovers reusable “skills” as low-rank directions in parameter space. Each skill is a rank-R basis spanning a subspace of the full parameter space. A gating network selects which skills to activate based on the initial task gradient signal. During adaptation, only the skill coefficients are updated, enabling efficient reuse.
Algorithm:
Constructors
| Constructor | Summary |
|---|---|
DiscoRLAlgorithm(DiscoRLOptions<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) |