Cross Attention Network for Few-shot Classification

Abstract

Few-shot classification aims to recognize unlabeled samples from unseen classes given only few labeled samples. The unseen classes and low-data problem make few-shot classification very challenging. Many existing approaches extracted features from labeled and unlabeled samples independently, as a result, the features are not discriminative enough. In this work, we propose a novel Cross Attention Network to address the challenging problems in few-shot classification. Firstly, Cross Attention Module is introduced to deal with the problem of unseen classes. The module generates cross attention maps for each pair of class feature and query sample feature so as to highlight the target object regions, making the extracted feature more discriminative. Secondly, a transductive inference algorithm is proposed to alleviate the low-data problem, which iteratively utilizes the unlabeled query set to augment the support set, thereby making the class features more representative. Extensive experiments on two benchmarks show our method is a simple, effective and computationally efficient framework and outperforms the state-of-the-arts.

12 pages, 4 figures. NeurIPS 2019 (Accepted)

Cited by in corpus (43)

• An Ensemble of Epoch-wise Empirical Bayes for Few-shot Learning
• Interventional Few-Shot Learning
• Efficient Attention: Attention with Linear Complexities
• TOAN: Target-Oriented Alignment Network for Fine-Grained Image Categorization with Few Labeled Samples
• Hypercorrelation Squeeze for Few-Shot Segmentation
• Transductive Information Maximization For Few-Shot Learning
• Self-Augmentation: Generalizing Deep Networks to Unseen Classes for Few-Shot Learning
• Rectifying the Shortcut Learning of Background for Few-Shot Learning
• Meta-Learned Confidence for Few-shot Learning
• Cross-Domain Few-Shot Learning by Representation Fusion
• Generalizing Face Forgery Detection with High-frequency Features
• Few-Shot Incremental Learning with Continually Evolved Classifiers
• Explanation-Guided Training for Cross-Domain Few-Shot Classification
• Self-training for Few-shot Transfer Across Extreme Task Differences
• Learning Dynamic Alignment via Meta-filter for Few-shot Learning
• Few-Shot Action Recognition with Compromised Metric via Optimal Transport
• Dense Relation Distillation with Context-aware Aggregation for Few-Shot Object Detection
• LibFewShot: A Comprehensive Library for Few-shot Learning
• Meta-Learning across Meta-Tasks for Few-Shot Learning
• Are Few-Shot Learning Benchmarks too Simple ? Solving them without Task Supervision at Test-Time
• Region Comparison Network for Interpretable Few-shot Image Classification
• Few-shot Unsupervised Domain Adaptation with Image-to-class Sparse Similarity Encoding
• XtarNet: Learning to Extract Task-Adaptive Representation for Incremental Few-Shot Learning
• Rapid Neural Architecture Search by Learning to Generate Graphs from Datasets
• Few-Shot Action Localization without Knowing Boundaries
• Attentive Graph Neural Networks for Few-Shot Learning
• Concept Learners for Few-Shot Learning
• Few-Shot Learning with Intra-Class Knowledge Transfer
• ReMP: Rectified Metric Propagation for Few-Shot Learning
• CAFENet: Class-Agnostic Few-Shot Edge Detection Network
• Transductive Few-Shot Learning: Clustering is All You Need?
• How to trust unlabeled data? Instance Credibility Inference for Few-Shot Learning
• Extended Few-Shot Learning: Exploiting Existing Resources for Novel Tasks
• BTS-Net: Bi-directional Transfer-and-Selection Network For RGB-D Salient Object Detection
• Improving Few-shot Learning with Weakly-supervised Object Localization
• Geo-Spatiotemporal Features and Shape-Based Prior Knowledge for Fine-grained Imbalanced Data Classification
• Axiomatic Explanations for Visual Search, Retrieval, and Similarity Learning
• Stochastic Whitening Batch Normalization
• A Transductive Maximum Margin Classifier for Few-Shot Learning
• Unsupervised Few-Shot Action Recognition via Action-Appearance Aligned Meta-Adaptation
• Towards A Conceptually Simple Defensive Approach for Few-shot classifiers Against Adversarial Support Samples
• Cooperative Bi-path Metric for Few-shot Learning
• Few-shot Learning with LSSVM Base Learner and Transductive Modules