MetaPACOHOptions<T, TInput, TOutput>
Options & Configuration · AiDotNet.MetaLearning.Options
Configuration options for Meta-PACOH: Hierarchical PAC-Bayesian Meta-Learning with per-group prior variances.
Inheritance: ModelOptions → MetaPACOHOptions<T, TInput, TOutput>
Implements: IMetaLearnerOptions<T>
How It Works
Meta-PACOH extends PACOH by introducing a hierarchical Bayesian structure where different parameter groups (e.g., layers) have independently learned prior variances. This enables the algorithm to express that some parameter groups should be more tightly constrained to the meta-learned prior while others can vary freely.
The algorithm meta-learns both a shared prior mean μ_P and per-group prior log-variances {log(σ²_g)} to minimize a hierarchical PAC-Bayesian bound.
Constructors
| Constructor | Summary |
|---|---|
MetaPACOHOptions(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 Delta { get; set; } |
Confidence parameter δ for the PAC-Bayesian bound. |
double HyperPriorLogVar { get; set; } |
Hyper-prior log-variance controlling how much per-group variances can deviate. |
double InitialLogVariance { get; set; } |
Initial log-variance for all prior groups. |
double InnerLearningRate { get; set; } |
|
double KLCoefficient { get; set; } |
KL divergence coefficient for the hierarchical PAC-Bayesian bound. |
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 NumPriorGroups { get; set; } |
Number of parameter groups with independent prior variances. |
int? RandomSeed { get; set; } |
Methods
| Method | Summary |
|---|---|
IMetaLearnerOptions<T> Clone() |
|
bool IsValid() |