Characteristic kernels on structured domains excel in robotics and human action recognition

Somayeh Danafar; Arthur Gretton; Jürgen Schmidhuber

doi:10.1007/978-3-642-15880-3_23

Characteristic kernels on structured domains excel in robotics and human action recognition

Somayeh Danafar, Arthur Gretton, Jürgen Schmidhuber

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

11 Scopus citations

Abstract

Embedding probability distributions into a sufficiently rich (characteristic) reproducing kernel Hilbert space enables us to take higher order statistics into account. Characterization also retains effective statistical relation between inputs and outputs in regression and classification. Recent works established conditions for characteristic kernels on groups and semigroups. Here we study characteristic kernels on periodic domains, rotation matrices, and histograms. Such structured domains are relevant for homogeneity testing, forward kinematics, forward dynamics, inverse dynamics, etc. Our kernel-based methods with tailored characteristic kernels outperform previous methods on robotics problems and also on a widely used benchmark for recognition of human actions in videos. © 2010 Springer-Verlag Berlin Heidelberg.

Original language	English (US)
Title of host publication	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages	264-279
Number of pages	16
DOIs	https://doi.org/10.1007/978-3-642-15880-3_23
State	Published - Nov 5 2010
Externally published	Yes

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-642-15880-3_23

Cite this

@inproceedings{7d5bee500005425eb63c7957d2a4561d,

title = "Characteristic kernels on structured domains excel in robotics and human action recognition",

abstract = "Embedding probability distributions into a sufficiently rich (characteristic) reproducing kernel Hilbert space enables us to take higher order statistics into account. Characterization also retains effective statistical relation between inputs and outputs in regression and classification. Recent works established conditions for characteristic kernels on groups and semigroups. Here we study characteristic kernels on periodic domains, rotation matrices, and histograms. Such structured domains are relevant for homogeneity testing, forward kinematics, forward dynamics, inverse dynamics, etc. Our kernel-based methods with tailored characteristic kernels outperform previous methods on robotics problems and also on a widely used benchmark for recognition of human actions in videos. {\textcopyright} 2010 Springer-Verlag Berlin Heidelberg.",

author = "Somayeh Danafar and Arthur Gretton and J{\"u}rgen Schmidhuber",

note = "Generated from Scopus record by KAUST IRTS on 2022-09-14",

year = "2010",

month = nov,

day = "5",

doi = "10.1007/978-3-642-15880-3_23",

language = "English (US)",

isbn = "364215879X",

pages = "264--279",

booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

Characteristic kernels on structured domains excel in robotics and human action recognition. / Danafar, Somayeh; Gretton, Arthur; Schmidhuber, Jürgen.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 2010. p. 264-279.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Characteristic kernels on structured domains excel in robotics and human action recognition

AU - Danafar, Somayeh

AU - Gretton, Arthur

AU - Schmidhuber, Jürgen

N1 - Generated from Scopus record by KAUST IRTS on 2022-09-14

PY - 2010/11/5

Y1 - 2010/11/5

N2 - Embedding probability distributions into a sufficiently rich (characteristic) reproducing kernel Hilbert space enables us to take higher order statistics into account. Characterization also retains effective statistical relation between inputs and outputs in regression and classification. Recent works established conditions for characteristic kernels on groups and semigroups. Here we study characteristic kernels on periodic domains, rotation matrices, and histograms. Such structured domains are relevant for homogeneity testing, forward kinematics, forward dynamics, inverse dynamics, etc. Our kernel-based methods with tailored characteristic kernels outperform previous methods on robotics problems and also on a widely used benchmark for recognition of human actions in videos. © 2010 Springer-Verlag Berlin Heidelberg.

AB - Embedding probability distributions into a sufficiently rich (characteristic) reproducing kernel Hilbert space enables us to take higher order statistics into account. Characterization also retains effective statistical relation between inputs and outputs in regression and classification. Recent works established conditions for characteristic kernels on groups and semigroups. Here we study characteristic kernels on periodic domains, rotation matrices, and histograms. Such structured domains are relevant for homogeneity testing, forward kinematics, forward dynamics, inverse dynamics, etc. Our kernel-based methods with tailored characteristic kernels outperform previous methods on robotics problems and also on a widely used benchmark for recognition of human actions in videos. © 2010 Springer-Verlag Berlin Heidelberg.

UR - http://link.springer.com/10.1007/978-3-642-15880-3_23

UR - http://www.scopus.com/inward/record.url?scp=78049339612&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-15880-3_23

DO - 10.1007/978-3-642-15880-3_23

M3 - Conference contribution

SN - 364215879X

SP - 264

EP - 279

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -

Characteristic kernels on structured domains excel in robotics and human action recognition

Abstract

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this