Abstract:The task of Few-shot learning (FSL) aims to transfer the knowledge learned from base categories with sufficient labelled data to novel categories with scarce known information. It is currently an important research question and has great practical values in the real-world applications. Despite extensive previous efforts are made on few-shot learning tasks, we emphasize that most existing methods did not take into account the distributional shift caused by sample selection bias in the FSL scenario. Such a selection bias can induce spurious correlation between the semantic causal features, that are causally and semantically related to the class label, and the other non-causal features. Critically, the former ones should be invariant across changes in distributions, highly related to the classes of interest, and thus well generalizable to novel classes, while the latter ones are not stable to changes in the distribution. To resolve this problem, we propose a novel data augmentation strategy dubbed as PatchMix that can break this spurious dependency by replacing the patch-level information and supervision of the query images with random gallery images from different classes from the query ones. We theoretically show that such an augmentation mechanism, different from existing ones, is able to identify the causal features. To further make these features to be discriminative enough for classification, we propose Correlation-guided Reconstruction (CGR) and Hardness-Aware module for instance discrimination and easier discrimination between similar classes. Moreover, such a framework can be adapted to the unsupervised FSL scenario. The utility of our method is demonstrated on the state-of-the-art results consistently achieved on several benchmarks including mini ImageNet, tiered ImageNet, CIFAR-FS, CUB, Cars, Places and Plantae, in all settings of single-domain, cross-domain and unsupervised FSL. By studying the intra-variance property of learned features and visualizing the learned features, we further quantitatively and qualitatively show that such a promising result is due to the effectiveness in learning causal features.

Learning Primitive-aware Discriminative Representations for FSL

Learning Primitive-aware Discriminative Representations for Few-shot Learning

Compositional Few-Shot Recognition with Primitive Discovery and Enhancing

DMN4: Few-shot Learning Via Discriminative Mutual Nearest Neighbor Neural Network

Semantic-based Selection, Synthesis, and Supervision for Few-shot Learning

Local Spatial Alignment Network for Few-Shot Learning

Co-Learning for Few-Shot Learning

PatchMix Augmentation to Identify Causal Features in Few-shot Learning

Multi-Attention Based Visual-Semantic Interaction for Few-Shot Learning

Mine-Distill-Prototypes for Complete Few-Shot Class-Incremental Learning in Image Classification

Few-Shot Learning With Global Class Representations

Envisioning Class Entity Reasoning by Large Language Models for Few-shot Learning

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

A two-generation based method for few-shot learning with few-shot instance-level privileged information

A Closer Look at Few-Shot Classification with Many Novel Classes

MDFM: Multi-Decision Fusing Model for Few-Shot Learning

Learning to Affiliate: Mutual Centralized Learning for Few-shot Classification

AdarGCN: Adaptive Aggregation GCN for Few-Shot Learning

Multi-directional Knowledge Transfer for Few-Shot Learning

Revisiting Mid-Level Patterns for Cross-Domain Few-Shot Recognition.