Dataset shift adaptation with active queries

Devis Tuia*, Edoardo Pasolli, William J. Emery

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference paper

9 Citations (Scopus)

Abstract

In remote sensing image classification, it is commonly assumed that the distribution of the classes is stable over the entire image. This way, training pixels labeled by photointerpretation are assumed to be representative of the whole image. However, differences in distribution of the classes throughout the image make this assumption weak and a model built on a single area may be suboptimal when applied to the rest of the image. In this paper, we investigate the use of active learning to correct the shifts that may appear when training and test data do not come from the same distribution. Experiments are carried out on a VHR remote sensing classification scenario showing that active learning can effectively learn the covariance shift and provide robust solutions.

Original languageEnglish
Title of host publication2011 Joint Urban Remote Sensing Event, JURSE 2011 - Proceedings
Pages121-124
Number of pages4
ISBN (Electronic)9781424486571
DOIs
Publication statusPublished - 2 Jun 2011
Externally publishedYes
EventIEEE GRSS and ISPRS Joint Urban Remote Sensing Event, JURSE 2011 - Munich, Germany
Duration: 11 Apr 201113 Apr 2011

Publication series

Name2011 Joint Urban Remote Sensing Event, JURSE 2011 - Proceedings

Conference

ConferenceIEEE GRSS and ISPRS Joint Urban Remote Sensing Event, JURSE 2011
CountryGermany
CityMunich
Period11/04/1113/04/11

Fingerprint Dive into the research topics of 'Dataset shift adaptation with active queries'. Together they form a unique fingerprint.

Cite this