BMAMLOptions<T, TInput, TOutput>
Options & Configuration · AiDotNet.MetaLearning.Options
Configuration options for BMAML: Bayesian Model-Agnostic Meta-Learning (Yoon et al., NeurIPS 2018).
Inheritance: ModelOptions → BMAMLOptions<T, TInput, TOutput>
Implements: IMetaLearnerOptions<T>
How It Works
BMAML uses Stein Variational Gradient Descent (SVGD) to maintain a set of particles (parameter vectors) that approximate the posterior over task-specific parameters. Instead of a single point estimate (as in MAML), BMAML produces a particle ensemble for uncertainty-aware predictions.
Key equations:
Constructors
| Constructor | Summary |
|---|---|
BMAMLOptions(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; } |
Must stay null: this algorithm takes no plain gradient inner step (it adapts without one, or by its own update rule), so a supplied optimizer is rejected when the learner is built. |
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 InnerLearningRate { get; set; } |
|
double OuterLearningRate { get; set; } |
|
double ParticleInitScale { get; set; } |
Scale of Gaussian noise for initial particle perturbation from θ_0. |
double SVGDRepulsiveWeight { get; set; } |
Weight for the SVGD repulsive (entropy) term relative to the attractive (likelihood) term. |
double? GradientClipThreshold { get; set; } |
|
double? KernelBandwidth { get; set; } |
RBF kernel bandwidth. |
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 NumParticles { get; set; } |
Number of particles (ensemble members) for SVGD posterior approximation. |
int? RandomSeed { get; set; } |
Methods
| Method | Summary |
|---|---|
IMetaLearnerOptions<T> Clone() |
|
bool IsValid() |