Abstract:Few-shot Learning (FSL) aims to learn a classifier that can be easily adapted to recognize novel classes with only a few labeled examples. Recently, some works about FSL have yielded promising classification performance, where the image-level feature is used to calculate the similarity among samples for classification. However, the image-level feature ignores abundant fine-grained and structural information of objects that could be transferable and consistent between seen and unseen classes. How can humans easily identify novel classes with several samples? Some studies from cognitive science argue that humans recognize novel categories based on primitives. Although base and novel categories are non-overlapping, they share some primitives in common. Inspired by above research, we propose a Primitive Mining and Reasoning Network (PMRN) to learn primitive-aware representations based on metric-based FSL model. Concretely, we first add Self-supervision Jigsaw task (SSJ) for feature extractor parallelly, guiding the model encoding visual pattern corresponding to object parts into feature channels. Moreover, to mine discriminative representations, an Adaptive Channel Grouping (ACG) method is applied to cluster and weight spatially and semantically related visual patterns to generate a set of visual primitives. To further enhance the discriminability and transferability of primitives, we propose a visual primitive Correlation Reasoning Network (CRN) based on Graph Convolutional network to learn abundant structural information and internal correlation among primitives. Finally, a primitive-level metric is conducted for classification in a meta-task based on episodic training strategy. Extensive experiments show that our method achieves state-of-the-art results on miniImageNet and Caltech-UCSD Birds.

A Theory of Human-Like Few-Shot Learning

Learning from One and Only One Shot

Recent Advances of Few-Shot Learning Methods and Applications

Learning from Few Examples: A Summary of Approaches to Few-Shot Learning

A Review of Few ‐ shot Learning

Few-shot Learning for Multi-Modality Tasks

Learning Relations in Human-Like Style for Few-Shot Fine-Grained Image Classification

Human-like Few-Shot Learning via Bayesian Reasoning over Natural Language

Less is More: A Closer Look at Semantic-based Few-Shot Learning

Large Margin Few-Shot Learning

Low-shot Learning in Natural Language Processing

Gestalt-Guided Image Understanding for Few-Shot Learning

Human-level few-shot concept induction through minimax entropy learning

A Neocognitron Based on Multi-objective Optimization for Few-Shot Learning

Semantic-Based Few-Shot Learning by Interactive Psychometric Testing

Improving the Generalised Few-shot Learning by Semantic Information

Learn from Concepts: Towards the Purified Memory for Few-shot Learning.

Learning Primitive-aware Discriminative Representations for Few-shot Learning

Bayesian Inverse Graphics for Few-Shot Concept Learning

A General Theory for Training Learning Machine

Learning with few samples in deep learning for image classification, a mini-review