Structure-Aware Data Consolidation

Shihao Wu, Peter Bertholet, Hui Huang*, Daniel Cohen-Or, Minglun Gong, Matthias Zwicker

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

16 Scopus citations

Abstract

We present a structure-aware technique to consolidate noisy data, which we use as a pre-process for standard clustering and dimensionality reduction. Our technique is related to mean shift, but instead of seeking density modes, it reveals and consolidates continuous high density structures such as curves and surface sheets in the underlying data while ignoring noise and outliers. We provide a theoretical analysis under a Gaussian noise model, and show that our approach significantly improves the performance of many non-linear dimensionality reduction and clustering algorithms in challenging scenarios.

Original languageEnglish
Article number8046026
Pages (from-to)2529-2537
Number of pages9
JournalIEEE Transactions on Pattern Analysis and Machine Intelligence
Volume40
Issue number10
DOIs
StatePublished - 1 Oct 2018

Funding

FundersFunder number
National Natural Science Foundation of China61522213, 61379090
Natural Science Foundation of Shenzhen University827-000196
Guangdong Provincial Applied Science and Technology Research and Development Program2015A030312015
Natural Sciences and Engineering Research Council of Canadaunidentified, 2017-06086
Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung169151
NSFC-ISF61761146002, 2472/ 17

    Keywords

    • Data consolidation
    • clustering
    • dimensionality reduction
    • filtering
    • manifold denoising

    Fingerprint

    Dive into the research topics of 'Structure-Aware Data Consolidation'. Together they form a unique fingerprint.

    Cite this