FRNOptions<T, TInput, TOutput>

Options & Configuration · AiDotNet.MetaLearning.Options

Configuration options for FRN (Few-shot Classification via Feature Map Reconstruction) (Wertheimer et al., CVPR 2021).

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

Implements: IMetaLearnerOptions<T>

For Beginners

FRN asks “can I rebuild this query from the support set?”:

  1. For each class, collect support feature maps
  2. Try to reconstruct the query’s feature map using ONLY features from class k
  3. If reconstruction is good (low error): query likely belongs to class k
  4. If reconstruction is bad (high error): query likely NOT class k

It’s like asking: “Can I explain this new image using only the patterns I’ve seen from cats?” If yes, it’s probably a cat. If not, try the next class.

How It Works

FRN classifies queries by reconstructing their feature maps from class-specific support feature maps. The reconstruction error serves as the distance metric.

Reference: Wertheimer, D., Tang, L., & Hariharan, B. (2021). Few-Shot Classification With Feature Map Reconstruction Networks. CVPR 2021.

Constructors

Constructor Summary
FRNOptions(IFullModel<T, TInput, TOutput> metaModel) Initializes a new instance of FRNOptions.

Properties

Property Summary
IEpisodicDataLoader<T, TInput, TOutput> DataLoader { get; set; } Gets or sets the episodic data loader.
IFullModel<T, TInput, TOutput> MetaModel { get; set; } Gets or sets the feature extractor model.
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; }
double InnerLearningRate { get; set; }
double OuterLearningRate { get; set; }
double ReconstructionLambda { get; set; } Gets or sets the reconstruction regularization parameter (lambda).
double? GradientClipThreshold { get; set; }
int AdaptationSteps { get; set; } Number of gradient steps taken on each task’s support set (0: this algorithm adapts without gradient steps).
int CheckpointFrequency { get; set; }
int EvaluationFrequency { get; set; }
int EvaluationTasks { get; set; }
int MetaBatchSize { get; set; }
int NumComponents { get; set; } Gets or sets the number of reconstruction components per class.
int NumMetaIterations { get; set; }
int? RandomSeed { get; set; } Gets or sets the random seed.

Methods

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