Transfer Learning by Learning Projections from Target to Source

Antoine Cornuéjols*, Pierre-Alexandre Murena, Raphaël Olivier

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Abstract

Using transfer learning to help in solving a new classification task where labeled data is scarce is becoming popular. Numerous experiments with deep neural networks, where the representation learned on a source task is transferred to learn a target neural network, have shown the benefits of the approach. This paper, similarly, deals with hypothesis transfer learning. However, it presents a new approach where, instead of transferring a representation, the source hypothesis is kept and this is a translation from the target domain to the source domain that is learned. In a way, a change of representation is learned. We show how this method performs very well on a classification of time series task where the space of time series is changed between source and target.
Original languageEnglish
Title of host publicationAdvances in Intelligent Data Analysis XVIII
Subtitle of host publication18th International Symposium on Intelligent Data Analysis, IDA 2020, Konstanz, Germany, April 27–29, 2020, Proceedings
Pages119-131
ISBN (Electronic)978-3-030-44584-3
DOIs
Publication statusPublished - 27 Apr 2020
MoE publication typeA4 Article in a conference publication
EventInternational Symposium on Intelligent Data Analysis - Konstanz, Germany
Duration: 27 Apr 202029 Apr 2020
Conference number: 18

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume12080
ISSN (Print)0302-9743

Conference

ConferenceInternational Symposium on Intelligent Data Analysis
Abbreviated titleIDA
CountryGermany
CityKonstanz
Period27/04/202029/04/2020

Fingerprint Dive into the research topics of 'Transfer Learning by Learning Projections from Target to Source'. Together they form a unique fingerprint.

Cite this