Skip to main navigation Skip to search Skip to main content

Deep kernel video approximation for unsupervised action segmentation

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

Abstract

This work focuses on per-video unsupervised action segmentation, which is of interest to applications where storing large datasets is either not possible, or nor permitted. We propose to segment videos by learning in deep kernel space, to approximate the underlying frame distribution, as closely as possible. To define this closeness metric between the original video distribution and its approximation, we rely on maximum mean discrepancy (MMD) which is a geometry-preserving metric in distribution space, and thus gives more reliable estimates. Moreover, unlike the commonly used optimal transport metric, MMD is both easier to optimize, and faster. We choose to use neural tangent kernels (NTKs) to define the kernel space where MMD operates, because of their improved descriptive power as opposed to fixed kernels. And, also, because NTKs sidestep the trivial solution, when jointly learning the inputs (video approximation) and the kernel function. Finally, we show competitive results when compared to state-of-the-art per-video methods, on six standard benchmarks. Additionally, our method has higher scores than prior agglomerative work, when the number of segments is unknown.
Original languageEnglish
Title of host publicationPattern recognition
Subtitle of host publication28th International conference, ICPR 2026, Lyon, France, August 17–22, 2026, Proceedings, Part I
EditorsMaria De Marsico, Tin Kam Ho, Frederic Jurie, Cheng-Lin Liu, Daniel Lopresti, Ingela Nyström, Jean-Marc Ogier, Arun Ross, Liang Wang
PublisherSpringer Cham
Pages32-47
Number of pages16
Edition1st
ISBN (Electronic)9783032313355
ISBN (Print)9783032313348
DOIs
Publication statusPublished - Aug 2026
Event 28th International conference on pattern recognition - Lyon, France
Duration: 17 Aug 2026 → 22 Aug 2026
Conference number: 28th

Publication series

NameLecture notes in computer science
PublisherSpringer Nature
Volume16812
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference 28th International conference on pattern recognition
Abbreviated titleICPR 2026
Country/TerritoryFrance
CityLyon
Period17/08/26 → 22/08/26

Keywords

  • unsupervised action segmentation
  • MMD
  • infinite NTK

Fingerprint

Dive into the research topics of 'Deep kernel video approximation for unsupervised action segmentation'. Together they form a unique fingerprint.

Cite this