Abstract:Prototypical Network is a popular few-shot solver that aims at establishing a feature metric generalizable to novel few-shot classification (FSC) tasks using deep neural networks. However, its performance drops dramatically when generalizing to the FSC tasks in new domains. In this study, we revisit this problem and argue that the devil lies in the simplicity bias pitfall in neural networks. In specific, the network tends to focus on some biased shortcut features (e.g., color, shape, etc.) that are exclusively sufficient to distinguish very few classes in the meta-training tasks within a pre-defined domain, but fail to generalize across domains as some desirable semantic features. To mitigate this problem, we propose a Local-global Distillation Prototypical Network (LDP-net). Different from the standard Prototypical Network, we establish a two-branch network to classify the query image and its random local crops, respectively. Then, knowledge distillation is conducted among these two branches to enforce their class affiliation consistency. The rationale behind is that since such global-local semantic relationship is expected to hold regardless of data domains, the local-global distillation is beneficial to exploit some cross-domain transferable semantic features for feature metric establishment. Moreover, such local-global semantic consistency is further enforced among different images of the same class to reduce the intra-class semantic variation of the resultant feature. In addition, we propose to update the local branch as Exponential Moving Average (EMA) over training episodes, which makes it possible to better distill cross-episode knowledge and further enhance the generalization performance. Experiments on eight cross-domain FSC benchmarks empirically clarify our argument and show the state-of-the-art results of LDP-net. Code is available in https://github.com/NWPUZhoufei/LDP-Net

Boosting Few-Shot Image Recognition Via Domain Alignment Prototypical Networks

Enhancing Image Classification Using Few-Shot Learning Prototypical Networks with ResNet-18: Detection, Accuracy Enhancement, and Optimization

Cost-effective CNNs-based Prototypical Networks for Few-Shot Relation Classification Across Domains

Revisiting Prototypical Network for Cross Domain Few-Shot Learning

DCPNet: Distribution Calibration Prototypical Network for Few-Shot Image Classification

Multi-domain few-shot image recognition with knowledge transfer

Transductive prototypical network for few-shot classification

FP_AINet: Fusion Prototype with Adaptive Induction Network for Few-Shot Learning

PANet: Few-Shot Image Semantic Segmentation With Prototype Alignment

Boosting Few-Shot Object Detection with Discriminative Representation and Class Margin

Bidirectional Matching Prototypical Network for Few-Shot Image Classification

SSL-ProtoNet: Self-supervised Learning Prototypical Networks for few-shot learning

Domain Mapping Network for Remote Sensing Cross-Domain Few-Shot Classification

Boosting Few-Shot Segmentation via Instance-Aware Data Augmentation and Local Consensus Guided Cross Attention

Meta-Prototypical Learning for Domain-Agnostic Few-Shot Recognition

Attentive Prototype Few-Shot Learning with Capsule Network-Based Embedding

A Few-shot-learning-based Method to Object Recognition in Multiple Scenarios.

Attentive Matching Network for Few-Shot Learning.

Category Alignment Mechanism for Few-Shot Image Classification

Few-shot Image Classification Via Hybrid Representation

Attributes-Guided and Pure-Visual Attention Alignment for Few-Shot Recognition