An Information Retrieval Approach for Finding Dependent Subspaces of Multiple Views

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

Researchers

Research units

Abstract

Finding relationships between multiple views of data is essential both in exploratory analysis and as pre-processing for predictive tasks. A prominent approach is to apply variants of Canonical Correlation Analysis (CCA), a classical method seeking correlated components between views. The basic CCA is restricted to maximizing a simple dependency criterion, correlation, measured directly between data coordinates.We introduce a new method that finds dependent subspaces of views directly optimized for the data analysis task of neighbor retrieval between multiple views. We optimize mappings for each view such as linear transformations to maximize cross-view similarity between neighborhoods of data samples. The criterion arises directly from the well-defined retrieval task, detects nonlinear and local similarities, measures dependency of data relationships rather than only individual data coordinates, and is related to well understood measures of information retrieval quality. In experiments the proposed method outperforms alternatives in preserving cross-view neighborhood similarities, and yields insights into local dependencies between multiple views.

Details

Original languageEnglish
Title of host publicationMachine Learning and Data Mining in Pattern Recognition
Subtitle of host publication13th International Conference, MLDM 2017 New York, NY, USA, July 15 – 20, 2017, Proceedings
EditorsPetra Perner
Publication statusPublished - Jul 2017
MoE publication typeA4 Article in a conference publication
EventInternational Conference on Machine Learning and Data Mining - New York, United States
Duration: 15 Jul 201720 Jul 2017
Conference number: 13

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Number10358
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceInternational Conference on Machine Learning and Data Mining
Abbreviated titleMLDM
CountryUnited States
CityNew York
Period15/07/201720/07/2017

ID: 14269329