Uniform Sampling over Episode Difficulty

2021/08/03 by Sébastien M. R. Arnold, Guneet S. Dhillon, Arnold, Sébastien M. R. +5
Computer Science · #Artificial Intelligence (cs.AI) #Computer Vision and Pattern Recognition (cs.CV) #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning and Data Classification #Multimodal Machine Learning Applications

paper · pdf · doi:10.48550/arxiv.2108.01662

openalex publication_date 2021/08/03 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Episodic training is a core ingredient of few-shot learning to train models on tasks with limited labelled data. Despite its success, episodic training remains largely understudied, prompting us to ask the question: what is the best way to sample episodes? In this paper, we first propose a method to approximate episode sampling distributions based on their difficulty. Building on this method, we perform an extensive analysis and find that sampling uniformly over episode difficulty outperforms other sampling schemes, including curriculum and easy-/hard-mining. As the proposed sampling method is algorithm agnostic, we can leverage these insights to improve few-shot learning accuracies across many episodic training algorithms. We demonstrate the efficacy of our method across popular few-shot learning datasets, algorithms, network architectures, and protocols.

Citations

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• A Stochastic Approximation Method
• Accelerating Minibatch Stochastic Gradient Descent using Stratified Sampling
• An analysis of variance test for normality (complete samples)
• Prioritized Experience Replay
• Prototypical Networks for Few-shot Learning
• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• Recasting Gradient-Based Meta-Learning as Hierarchical Bayes
• On First-Order Meta-Learning Algorithms
• Meta-learning with differentiable closed-form solvers
• DropBlock: A regularization method for convolutional networks
• Meta-Learning with Implicit Gradients
• Revisiting Meta-Learning as Supervised Learning
• Generalizing from a Few Examples
• Expert Training: Task Hardness Aware Meta-Learning for Few-Shot Classification
• On Episodes, Prototypical Networks, and Few-shot Learning
• When Do Curricula Work?
• Optimization Methods for Large-Scale Machine Learning

Related

• On Episodes, Prototypical Networks, and Few-shot Learning
• Episode-specific Fine-tuning for Metric-based Few-shot Learners with Optimization-based Training
• Enhancing Few-Shot Classification without Forgetting through Multi-Level Contrastive Constraints
• Episodic-free Task Selection for Few-shot Learning
• Few-Shot Learning as Domain Adaptation: Algorithm and Analysis