OpenMAMLPlusOptions<T, TInput, TOutput>

Options & Configuration ยท AiDotNet.MetaLearning.Options

Configuration options for Open-MAML++: MAML extended for open-set recognition with novelty detection.

Inheritance: ModelOptions โ†’ OpenMAMLPlusOptions<T, TInput, TOutput>

Implements: IMetaLearnerOptions<T>

How It Works

Open-MAML++ extends MAML to handle open-set scenarios where test tasks may contain classes not seen during meta-training. It meta-learns per-parameter learning rates (like MAML++) and a novelty detection threshold based on prediction entropy. During adaptation, predictions with entropy above the threshold are flagged as novel/unknown.

Constructors

Constructor Summary
OpenMAMLPlusOptions(IFullModel<T, TInput, TOutput> metaModel) Creates options for meta-learning metaModel, with the default settings.

Properties

Property Summary
IEpisodicDataLoader<T, TInput, TOutput> DataLoader { get; set; } Episodic data loader that supplies the meta-training tasks; null when tasks are passed to training directly.
IFullModel<T, TInput, TOutput> MetaModel { get; set; } The model whose parameters the meta-learner trains to adapt quickly to new tasks.
IGradientBasedOptimizer<T, TInput, TOutput> InnerOptimizer { get; set; } Optimizer for task adaptation.
IGradientBasedOptimizer<T, TInput, TOutput> MetaOptimizer { get; set; } Optimizer for the meta-update.
ILossFunction<T> LossFunction { get; set; } Loss function for meta-training; null uses the default multi-class classification loss.
bool EnableCheckpointing { get; set; }
bool LearnPerParamLR { get; set; } Whether to meta-learn per-parameter learning rates (MAML++ style).
bool UseFirstOrder { get; set; } Whether to use the first-order approximation of the meta-gradient.
double EntropyRegWeight { get; set; } Weight for the entropy regularization loss.
double InitialNoveltyThreshold { get; set; } Initial novelty threshold (entropy-based).
double InnerLearningRate { get; set; }
double MultiStepLossWeight { get; set; } Multi-step loss coefficient: weight for intermediate adaptation step losses.
double OuterLearningRate { get; set; }
double? GradientClipThreshold { get; set; }
int AdaptationSteps { get; set; }
int CheckpointFrequency { get; set; }
int EvaluationFrequency { get; set; }
int EvaluationTasks { get; set; }
int MetaBatchSize { get; set; }
int NumMetaIterations { get; set; }
int? RandomSeed { get; set; }

Methods

Method Summary
IMetaLearnerOptions<T> Clone()
bool IsValid()