Improving Few-Shot Learning with Auxiliary Self-Supervised Pretext Tasks

Abstract

Recent work on few-shot learning \cite{tian2020rethinking} showed that quality of learned representations plays an important role in few-shot classification performance. On the other hand, the goal of self-supervised learning is to recover useful semantic information of the data without the use of class labels. In this work, we exploit the complementarity of both paradigms via a multi-task framework where we leverage recent self-supervised methods as auxiliary tasks. We found that combining multiple tasks is often beneficial, and that solving them simultaneously can be done efficiently. Our results suggest that self-supervised auxiliary tasks are effective data-dependent regularizers for representation learning. Our code is available at: \url{https://github.com/nathanielsimard/improving-fs-ssl}.

Research project report for graduate class IFT 6268-A2020 on Self-supervised Representation Learning

References in corpus (14)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Prototypical Networks for Few-shot Learning
• Improved Regularization of Convolutional Neural Networks with Cutout
• Improved Baselines with Momentum Contrastive Learning
• Theoretical Models of Learning to Learn
• Data-Efficient Image Recognition with Contrastive Predictive Coding
• Contrastive Multiview Coding
• DropBlock: A regularization method for convolutional networks
• Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need?
• Dynamic Few-Shot Visual Learning without Forgetting
• Data-Efficient Reinforcement Learning with Self-Predictive Representations
• Understanding Self-supervised Learning with Dual Deep Networks
• Image Enhanced Rotation Prediction for Self-Supervised Learning