Transfer learning based few-shot classification using optimal transport mapping from preprocessed latent space of backbone neural network

2021/02/09 by Tomáš Chobola, Chobola, Tomáš, Daniel Vašata +3
Computer Science · #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 ELM #Multimodal Machine Learning Applications #cs.CV #cs.LG

paper · pdf · doi:10.48550/arxiv.2102.05176

openalex publication_date 2021/02/09 · arxiv created 2021/02/11 · arxiv updated 2021/02/12 · openalex created_date 2021/02/15 · openalex updated_date 2026/07/28

Abstract

MetaDL Challenge 2020 focused on image classification tasks in few-shot settings. This paper describes second best submission in the competition. Our meta learning approach modifies the distribution of classes in a latent space produced by a backbone network for each class in order to better follow the Gaussian distribution. After this operation which we call Latent Space Transform algorithm, centers of classes are further aligned in an iterative fashion of the Expectation Maximisation algorithm to utilize information in unlabeled data that are often provided on top of few labelled instances. For this task, we utilize optimal transport mapping using the Sinkhorn algorithm. Our experiments show that this approach outperforms previous works as well as other variants of the algorithm, using K-Nearest Neighbour algorithm, Gaussian Mixture Models, etc.

Citations

• Wide Residual Networks
• Prototypical Networks for Few-shot Learning
• Meta-learning with differentiable closed-form solvers
• Generalizing from a Few Examples

Related

• Squeezing Backbone Feature Distributions to the Max for Efficient Few-Shot Learning
• Leveraging the Feature Distribution in Transfer-based Few-Shot Learning
• Few-Shot Learning for Image Classification of Common Flora
• Self-Supervised Prototypical Transfer Learning for Few-Shot Classification
• Few-shot Image Classification: Just Use a Library of Pre-trained Feature Extractors and a Simple Classifier