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.

Target-oriented Few-shot Transferring Via Measuring Task Similarity

Reweighting and Information-Guidance Networks for Few-Shot Learning

Local Spatial Alignment Network for Few-Shot Learning

Pixel Matching Network for Cross-Domain Few-Shot Segmentation

TAFSSL: Task-Adaptive Feature Sub-Space Learning for few-shot classification

Hybrid Consistency Training with Prototype Adaptation for Few-Shot Learning

Transductive Few-shot Learning with Prototype-based Label Propagation by Iterative Graph Refinement

Prototype Completion for Few-Shot Learning

SpatialFormer: Semantic and Target Aware Attentions for Few-Shot Learning

Self-taught cross-domain few-shot learning with weakly supervised object localization and task-decomposition

Balancing Feature Alignment and Uniformity for Few-Shot Classification.

Cross-Domain Cross-Set Few-Shot Learning via Learning Compact and Aligned Representations

Tailoring Embedding Function to Heterogeneous Few-Shot Tasks by Global and Local Feature Adaptors

Learning task-specific discriminative embeddings for few-shot image classification

A Closer Look at Few-Shot Video Classification: A New Baseline and Benchmark

Multi-directional Knowledge Transfer for Few-Shot Learning

Interventional Few-Shot Learning

Task-aware Part Mining Network for Few-Shot Learning

Few-Shot Keypoint Detection as Task Adaptation via Latent Embeddings

Task-Related Saliency for Few-Shot Image Classification

Task Adaptive Feature Distribution Based Network for Few-shot Fine-grained Target Classification