• Corpus ID: 235185409

Meta-Learning for One-Class Classification with Few Examples using Order-Equivariant Network

  title={Meta-Learning for One-Class Classification with Few Examples using Order-Equivariant Network},
  author={Ademola Oladosu and Tony Xu and Philip Ekfeldt and Brian A. Kelly and M. Cranmer and Shirley Ho and Adrian. M. Price-Whelan and Gabriella Contardo},
This paper presents a meta-learning framework for few-shots One-Class Classification (OCC) at test-time, a setting where labeled examples are only available for the positive class, and no supervision is given for the negative example. We consider that we have a set of ‘one-class classification’ objective-tasks with only a small set of positive examples available for each task, and a set of training tasks with full supervision (i.e. highly imbalanced classification). We propose an approach using… 

Figures and Tables from this paper

Spurious correlations between galaxies and multi-epoch image stacks in the DESI Legacy Surveys

A non-negligible source of systematic bias in cosmological analyses of galaxy surveys is the on-sky modulation caused by foregrounds and variable image characteristics such as observing conditions.



Meta Learning for Few-Shot One-class Classification

A method that can perform one-class classification given only a small number of examples from the target class and none from the others is proposed, and it is shown how the Support Vector Data Description method can be used with this method, and a simpler variant based on Prototypical Networks that obtains comparable performance is proposed.

Meta-Learning Deep Energy-Based Memory Models

A novel meta-learning approach to energy-based memory models (EBMM) that allows one to use an arbitrary neural architecture as an energy model and quickly store patterns in its weights and is capable of associative retrieval that outperforms existing memory systems in terms of the reconstruction error and compression rate.

Deep Learning for Anomaly Detection: A Survey

This survey presents a structured and comprehensive overview of research methods in deep learning-based anomaly detection, grouping state-of-the-art deep anomaly detection research techniques into different categories based on the underlying assumptions and approach adopted.

Learning to learn by gradient descent by gradient descent

This paper shows how the design of an optimization algorithm can be cast as a learning problem, allowing the algorithm to learn to exploit structure in the problems of interest in an automatic way.

Stellar Streams and Clouds in the Galactic Halo

Recent years have seen the discovery of an ever growing number of stellar debris streams and clouds. These structures are typically detected as extended and often curvilinear overdensities of

Learning classifiers from only positive and unlabeled data

This paper shows that models trained using the new methods perform better than the current state-of-the-art biased SVM method for learning from positive and unlabeled examples, and applies them to solve a real-world problem: identifying protein records that should be included in an incomplete specialized molecular biology database.

Probing the nature of dark matter particles with stellar streams

A key prediction of the standard cosmological model—which relies on the assumption that dark matter is cold, i.e. non-relativistic at the epoch of structure formation—is the existence of a large

Learning to Compare: Relation Network for Few-Shot Learning

A conceptually simple, flexible, and general framework for few-shot learning, where a classifier must learn to recognise new classes given only few examples from each, which is easily extended to zero- shot learning.

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

We propose an algorithm for meta-learning that is model-agnostic, in the sense that it is compatible with any model trained with gradient descent and applicable to a variety of different learning

Self-labeled techniques for semi-supervised learning: taxonomy, software and empirical study

This paper provides a survey of self-labeled methods for semi-supervised classification and proposes a taxonomy based on the main characteristics presented in them, aiming to measure their performance in terms of transductive and inductive classification capabilities.