MetaDMOptions<T, TInput, TOutput>

Options & Configuration · AiDotNet.MetaLearning.Options

Configuration options for Meta-DM: Applications of Diffusion Models on Few-Shot Learning (Hu et al., ICIP 2024).

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

Implements: IMetaLearnerOptions<T>

How It Works

Meta-DM uses a DDPM-style diffusion model as a data augmentation module for few-shot learning. It generates synthetic support samples conditioned on the existing few-shot support set, then trains on the enriched dataset. This is a modular augmentation strategy composable with any gradient-based meta-learning algorithm.

Constructors

Constructor Summary
MetaDMOptions(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 regression 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 noise schedule parameter.
double BetaStart { get; set; } Starting noise schedule parameter.
double InnerLearningRate { get; set; }
double MatchingWeight { get; set; } Weight of the distribution matching loss relative to the task loss.
double OuterLearningRate { get; set; }
double? GradientClipThreshold { get; set; }
int AdaptationSteps { get; set; }
int CheckpointFrequency { get; set; }
int DenoisingSteps { get; set; } Number of denoising steps for generation (≤ DiffusionTimesteps).
int DiffusionTimesteps { get; set; } Number of diffusion timesteps for generation.
int EvaluationFrequency { get; set; }
int EvaluationTasks { get; set; }
int MetaBatchSize { get; set; }
int NumMetaIterations { get; set; }
int PrototypeDim { get; set; } Dimensionality of the prototype embeddings for distribution matching.
int SyntheticSamplesPerClass { get; set; } Number of synthetic samples to generate per class for augmentation.
int? RandomSeed { get; set; }

Methods

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