MetaDDPMAlgorithm<T, TInput, TOutput>
Models & Types · AiDotNet.MetaLearning.Algorithms
Implementation of Meta-DDPM: Meta-Learning with Denoising Diffusion Probabilistic Models.
Inheritance: ModelBase<T, TInput, TOutput> → MetaLearnerBase<T, TInput, TOutput> → MetaDDPMAlgorithm<T, TInput, TOutput>
Implements: IGeneratedParameterRegistrar<T>
How It Works
Meta-DDPM meta-learns a denoising diffusion model that generates task-specific model weights conditioned on support set embeddings. It uses the full DDPM framework with a linear noise schedule and EMA parameter averaging for stable generation.
Algorithm:
Key difference from MetaDiff: Meta-DDPM uses the standard DDPM sampling algorithm with full variance (σ_t = √β_t), while MetaDiff uses the deterministic DDIM-style denoising. Meta-DDPM also includes EMA for denoiser parameter stability.
Constructors
| Constructor | Summary |
|---|---|
MetaDDPMAlgorithm(MetaDDPMOptions<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) |