Abstract:Few-shot learning (FSL) aims to recognize novel classes with few examples. Pre-training based methods effectively tackle the problem by pre-training a feature extractor and then fine-tuning it through the nearest centroid based meta-learning. However, results show that the fine-tuning step makes marginal improvements. In this paper, 1) we figure out the reason, i.e., in the pre-trained feature space, the base classes already form compact clusters while novel classes spread as groups with large variances, which implies that fine-tuning feature extractor is less meaningful; 2) instead of fine-tuning feature extractor, we focus on estimating more representative prototypes. Consequently, we propose a novel prototype completion based meta-learning framework. This framework first introduces primitive knowledge (i.e., class-level part or attribute annotations) and extracts representative features for seen attributes as priors. Second, a part/attribute transfer network is designed to learn to infer the representative features for unseen attributes as supplementary priors. Finally, a prototype completion network is devised to learn to complete prototypes with these priors. Moreover, to avoid the prototype completion error, we further develop a Gaussian based prototype fusion strategy that fuses the mean-based and completed prototypes by exploiting the unlabeled samples. At last, we also develop an economic prototype completion version for FSL, which does not need to collect primitive knowledge, for a fair comparison with existing FSL methods without external knowledge. Extensive experiments show that our method: (i) obtains more accurate prototypes; (ii) achieves superior performance on both inductive and transductive FSL settings. Our codes are open-sourced at https://github.com/zhangbq-research/Prototype_Completion_for_FSL.

ProtoRefine: Enhancing Prototypes with Similar Structure in Few-Shot Learning

Feature Transformation for Few-Shot Learning

Learning Class-level Prototypes for Few-shot Learning

PrototypeFormer: Learning to Explore Prototype Relationships for Few-shot Image Classification

Prototype Completion with Primitive Knowledge for Few-Shot Learning

Self-Promoted Prototype Refinement for Few-Shot Class-Incremental Learning

Reweighting and Information-Guidance Networks for Few-Shot Learning

Prototype Completion for Few-Shot Learning

Prototype Rectification with Region-Wise Foreground Enhancement for Few-Shot Classification.

Membership-Grade Based Prototype Rectification for Fine-Grained Few-Shot Classification

Task-aware prototype refinement for improved few-shot learning

Complementary Features Based Prototype Self-Updating for Few-Shot Learning

Prototype Relationship Optimization Network for Few‐Shot Learning

Few-shot learning with representative global prototype

Meta-Learning Based Prototype-Relation Network for Few-Shot Classification

Query-Guided Prototype Optimization for Few-Shot Classification

Few-shot Relation Classification Using Clustering-Based Prototype Modification.

Few-Shot Segmentation via Complementary Prototype Learning and Cascaded Refinement.

Improved prototypical network for active few-shot learning

Adaptive Prototype Few-Shot Image Classification Method Based on Feature Pyramid

Hybrid Enhancement-based prototypical networks for few-shot relation classification