MetaDiffOptions<T, TInput, TOutput>

Options & Configuration · AiDotNet.MetaLearning.Options

Configuration options for MetaDiff: Meta-Learning with Conditional Diffusion for Few-Shot Learning (Zhang et al., AAAI 2024).

Inheritance: ModelOptions → MetaDiffOptions<T, TInput, TOutput>

Implements: IMetaLearnerOptions<T>

How It Works

MetaDiff reframes the inner-loop gradient descent of meta-learning as a reverse diffusion process over model weights. A task-conditional denoising network iteratively removes noise to produce task-specific weights, conditioned on support-set features.

Constructors

Constructor Summary
MetaDiffOptions(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 UseFirstOrder { get; set; } Whether to use the first-order approximation of the meta-gradient.
double BetaEnd { get; set; } Ending value of the linear noise schedule β_T.
double BetaStart { get; set; } Starting value of the linear noise schedule β_1.
double InnerLearningRate { get; set; }
double OuterLearningRate { get; set; } Learning rate of the outer (meta) update.
double? GradientClipThreshold { get; set; }
int AdaptationSteps { get; set; }
int CheckpointFrequency { get; set; }
int DiffusionSteps { get; set; } Number of diffusion timesteps for the full forward/reverse process.
int EvaluationFrequency { get; set; }
int EvaluationTasks { get; set; }
int MetaBatchSize { get; set; }
int NumMetaIterations { get; set; }
int SamplingSteps { get; set; } Number of denoising steps used during inference (≤ DiffusionSteps).
int TaskConditionDim { get; set; } Dimensionality of the task conditioning vector computed from support features.
int? RandomSeed { get; set; }

Methods

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