MetaBaselineOptions<T, TInput, TOutput>
Options & Configuration · AiDotNet.MetaLearning.Options
Configuration options for Meta-Baseline (Chen et al., ICLR 2021).
Inheritance: ModelOptions → MetaBaselineOptions<T, TInput, TOutput>
Implements: IMetaLearnerOptions<T>
For Beginners
Meta-Baseline is a “simple but strong” approach:
Phase 1: Pre-training Train a classifier normally on all base classes (standard cross-entropy). This produces a strong feature extractor.
Phase 2: Meta-training Fine-tune with episodic training using cosine similarity to class centroids. This adapts the features for few-shot nearest-centroid classification.
Why it’s effective: A well-pretrained backbone provides excellent features. The meta-training phase just polishes them for the nearest-centroid setting.
How It Works
Meta-Baseline first pre-trains a feature extractor with standard classification, then meta-trains with cosine-similarity-based nearest-centroid classification. This two-phase approach provides a strong baseline that outperforms many complex methods.
Reference: Chen, Y., Liu, Z., Xu, H., Darrell, T., & Wang, X. (2021). Meta-Baseline: Exploring Simple Meta-Learning for Few-Shot Learning. ICLR 2021.
Constructors
| Constructor | Summary |
|---|---|
MetaBaselineOptions(IFullModel<T, TInput, TOutput> metaModel) |
Initializes a new instance of MetaBaselineOptions. |
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 Temperature { get; set; } |
Gets or sets the temperature for cosine similarity scaling. |
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 NumMetaIterations { get; set; } |
|
int? RandomSeed { get; set; } |
Gets or sets the random seed. |
Methods
| Method | Summary |
|---|---|
IMetaLearnerOptions<T> Clone() |
|
bool IsValid() |