SimpleShot: Revisiting Nearest-Neighbor Classification for Few-Shot Learning

Abstract

Few-shot learners aim to recognize new object classes based on a small number of labeled training examples. To prevent overfitting, state-of-the-art few-shot learners use meta-learning on convolutional-network features and perform classification using a nearest-neighbor classifier. This paper studies the accuracy of nearest-neighbor baselines without meta-learning. Surprisingly, we find simple feature transformations suffice to obtain competitive few-shot learning accuracies. For example, we find that a nearest-neighbor classifier used in combination with mean-subtraction and L2-normalization outperforms prior results in three out of five settings on the miniImageNet dataset.

References in corpus (11)

• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• Prototypical Networks for Few-shot Learning
• Matching Networks for One Shot Learning
• A Simple Neural Attentive Meta-Learner
• Meta-SGD: Learning to Learn Quickly for Few-Shot Learning
• Few-Shot Learning with Graph Neural Networks
• TADAM: Task dependent adaptive metric for improved few-shot learning
• The Devil is in the Tails: Fine-grained Classification in the Wild
• Few-Shot Learning via Embedding Adaptation with Set-to-Set Functions
• Few-Shot Learning Through an Information Retrieval Lens
• Discriminative k-shot learning using probabilistic models

Cited by in corpus (41)

• Interventional Few-Shot Learning
• Representation Learning for Natural Language Processing
• Few-Shot Learning via Embedding Adaptation with Set-to-Set Functions
• Learning to detect an animal sound from five examples
• Multi-level Second-order Few-shot Learning
• Identifying and Compensating for Feature Deviation in Imbalanced Deep Learning
• Transductive Information Maximization For Few-Shot Learning
• Enhancing Few-Shot Image Classification through Learnable Multi-Scale Embedding and Attention Mechanisms
• Leveraging the Feature Distribution in Transfer-based Few-Shot Learning
• Robust Saliency-Aware Distillation for Few-shot Fine-grained Visual Recognition
• A Closer Look at Few-Shot 3D Point Cloud Classification
• Self-training for Few-shot Transfer Across Extreme Task Differences
• On Episodes, Prototypical Networks, and Few-shot Learning
• Simple and Effective Few-Shot Named Entity Recognition with Structured Nearest Neighbor Learning
• Meta-Learning with Graph Neural Networks: Methods and Applications
• Learning Adaptive Classifiers Synthesis for Generalized Few-Shot Learning
• Few-Shot Learning with Uncertainty-based Quadruplet Selection for Interference Classification in GNSS Data
• Are Few-Shot Learning Benchmarks too Simple ? Solving them without Task Supervision at Test-Time
• Performance Analysis of Few-Shot Learning Approaches for Bangla Handwritten Character and Digit Recognition
• Modeling and Optimization Trade-off in Meta-learning
• Graph-based Interpolation of Feature Vectors for Accurate Few-Shot Classification
• Machine learning with limited data
• Structure-Enhanced Meta-Learning For Few-Shot Graph Classification
• Revisiting Pose-Normalization for Fine-Grained Few-Shot Recognition
• Joint Image-Instance Spatial-Temporal Attention for Few-shot Action Recognition
• Design Environment of Quantization-Aware Edge AI Hardware for Few-Shot Learning
• Aligning Visual Prototypes with BERT Embeddings for Few-Shot Learning
• Gaussian Process Meta Few-shot Classifier Learning via Linear Discriminant Laplace Approximation
• Predicting the Accuracy of a Few-Shot Classifier
• Few-shot Intent Classification and Slot Filling with Retrieved Examples
• Model-Agnostic Graph Regularization for Few-Shot Learning
• Transductive Few-Shot Learning: Clustering is All You Need?
• Generalized Few-Shot Video Classification with Video Retrieval and Feature Generation
• Fuzzy Simplicial Networks: A Topology-Inspired Model to Improve Task Generalization in Few-shot Learning
• Mutual-Information Based Few-Shot Classification
• Procrustean Training for Imbalanced Deep Learning
• DecisiveNets: Training Deep Associative Memories to Solve Complex Machine Learning Problems
• Reinforced Attention for Few-Shot Learning and Beyond
• How Sensitive are Meta-Learners to Dataset Imbalance?
• Learning by Examples Based on Multi-level Optimization
• Towards Enabling Meta-Learning from Target Models