Opening the black box: Revealing interpretable sequence motifs in kernel-based learning algorithms

Marina M.C. Vidovic; Nico Görnitz; Klaus Robert Müller; Gunnar Rätsch; Marius Kloft

doi:10.1007/978-3-319-23525-7_9

Opening the black box: Revealing interpretable sequence motifs in kernel-based learning algorithms

Marina M.C. Vidovic, Nico Görnitz, Klaus Robert Müller, Gunnar Rätsch, Marius Kloft

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

5 Citations (Scopus)

Abstract

This work is in the context of kernel-based learning algorithms for sequence data. We present a probabilistic approach to automatically extract, from the output of such string-kernel-based learning algorithms, the subsequences—or motifs—truly underlying the machine’s predictions. The proposed framework views motifs as free parameters in a probabilistic model, which is solved through a global optimization approach. In contrast to prevalent approaches, the proposed method can discover even difficult, long motifs, and could be combined with any kernel-based learning algorithm that is based on an adequate sequence kernel. We show that, by using a discriminate kernel machine such as a support vector machine, the approach can reveal discriminative motifs underlying the kernel predictor. We demonstrate the efficacy of our approach through a series of experiments on synthetic and real data, including problems from handwritten digit recognition and a large-scale human splice site data set from the domain of computational biology.

Original language	English
Title of host publication	Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2015
Editors	Vitor Santos Costa, Carlos Soares, Annalisa Appice, Annalisa Appice, Pedro Pereira Rodrigues, Vitor Santos Costa, Carlos Soares, João Gama, Alípio Jorge, Pedro Pereira Rodrigues, João Gama, Vitor Santos Costa, Alípio Jorge, Annalisa Appice, Pedro Pereira Rodrigues, João Gama, Annalisa Appice, Carlos Soares, Alípio Jorge, João Gama, Pedro Pereira Rodrigues, Vitor Santos Costa, Carlos Soares, Alípio Jorge
Publisher	Springer Verlag
Pages	137-153
Number of pages	17
ISBN (Print)	9783319235240, 9783319235240, 9783319235240, 9783319235240
DOIs	https://doi.org/10.1007/978-3-319-23525-7_9
Publication status	Published - 2015
Event	European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2015 - Porto, Portugal Duration: 2015 Sept 7 → 2015 Sept 11

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	9285
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2015
Country/Territory	Portugal
City	Porto
Period	15/9/7 → 15/9/11

Bibliographical note

Publisher Copyright:
© Springer International Publishing Switzerland 2015.

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/978-3-319-23525-7_9

Cite this

Vidovic, M. M. C., Görnitz, N., Müller, K. R., Rätsch, G., & Kloft, M. (2015). Opening the black box: Revealing interpretable sequence motifs in kernel-based learning algorithms. In V. S. Costa, C. Soares, A. Appice, A. Appice, P. P. Rodrigues, V. S. Costa, C. Soares, J. Gama, A. Jorge, P. P. Rodrigues, J. Gama, V. S. Costa, A. Jorge, A. Appice, P. P. Rodrigues, J. Gama, A. Appice, C. Soares, A. Jorge, J. Gama, P. P. Rodrigues, V. S. Costa, C. Soares, ... A. Jorge (Eds.), Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2015 (pp. 137-153). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9285). Springer Verlag. https://doi.org/10.1007/978-3-319-23525-7_9

Opening the black box: Revealing interpretable sequence motifs in kernel-based learning algorithms. / Vidovic, Marina M.C.; Görnitz, Nico; Müller, Klaus Robert et al.
Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2015. ed. / Vitor Santos Costa; Carlos Soares; Annalisa Appice; Annalisa Appice; Pedro Pereira Rodrigues; Vitor Santos Costa; Carlos Soares; João Gama; Alípio Jorge; Pedro Pereira Rodrigues; João Gama; Vitor Santos Costa; Alípio Jorge; Annalisa Appice; Pedro Pereira Rodrigues; João Gama; Annalisa Appice; Carlos Soares; Alípio Jorge; João Gama; Pedro Pereira Rodrigues; Vitor Santos Costa; Carlos Soares; Alípio Jorge. Springer Verlag, 2015. p. 137-153 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9285).

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

Vidovic, MMC, Görnitz, N, Müller, KR, Rätsch, G & Kloft, M 2015, Opening the black box: Revealing interpretable sequence motifs in kernel-based learning algorithms. in VS Costa, C Soares, A Appice, A Appice, PP Rodrigues, VS Costa, C Soares, J Gama, A Jorge, PP Rodrigues, J Gama, VS Costa, A Jorge, A Appice, PP Rodrigues, J Gama, A Appice, C Soares, A Jorge, J Gama, PP Rodrigues, VS Costa, C Soares & A Jorge (eds), Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2015. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9285, Springer Verlag, pp. 137-153, European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2015, Porto, Portugal, 15/9/7. https://doi.org/10.1007/978-3-319-23525-7_9

Vidovic MMC, Görnitz N, Müller KR, Rätsch G, Kloft M. Opening the black box: Revealing interpretable sequence motifs in kernel-based learning algorithms. In Costa VS, Soares C, Appice A, Appice A, Rodrigues PP, Costa VS, Soares C, Gama J, Jorge A, Rodrigues PP, Gama J, Costa VS, Jorge A, Appice A, Rodrigues PP, Gama J, Appice A, Soares C, Jorge A, Gama J, Rodrigues PP, Costa VS, Soares C, Jorge A, editors, Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2015. Springer Verlag. 2015. p. 137-153. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-23525-7_9

Vidovic, Marina M.C. ; Görnitz, Nico ; Müller, Klaus Robert et al. / Opening the black box : Revealing interpretable sequence motifs in kernel-based learning algorithms. Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2015. editor / Vitor Santos Costa ; Carlos Soares ; Annalisa Appice ; Annalisa Appice ; Pedro Pereira Rodrigues ; Vitor Santos Costa ; Carlos Soares ; João Gama ; Alípio Jorge ; Pedro Pereira Rodrigues ; João Gama ; Vitor Santos Costa ; Alípio Jorge ; Annalisa Appice ; Pedro Pereira Rodrigues ; João Gama ; Annalisa Appice ; Carlos Soares ; Alípio Jorge ; João Gama ; Pedro Pereira Rodrigues ; Vitor Santos Costa ; Carlos Soares ; Alípio Jorge. Springer Verlag, 2015. pp. 137-153 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{9337ef087c534bfba91cd09737674dae,

title = "Opening the black box: Revealing interpretable sequence motifs in kernel-based learning algorithms",

abstract = "This work is in the context of kernel-based learning algorithms for sequence data. We present a probabilistic approach to automatically extract, from the output of such string-kernel-based learning algorithms, the subsequences—or motifs—truly underlying the machine{\textquoteright}s predictions. The proposed framework views motifs as free parameters in a probabilistic model, which is solved through a global optimization approach. In contrast to prevalent approaches, the proposed method can discover even difficult, long motifs, and could be combined with any kernel-based learning algorithm that is based on an adequate sequence kernel. We show that, by using a discriminate kernel machine such as a support vector machine, the approach can reveal discriminative motifs underlying the kernel predictor. We demonstrate the efficacy of our approach through a series of experiments on synthetic and real data, including problems from handwritten digit recognition and a large-scale human splice site data set from the domain of computational biology.",

author = "Vidovic, {Marina M.C.} and Nico G{\"o}rnitz and M{\"u}ller, {Klaus Robert} and Gunnar R{\"a}tsch and Marius Kloft",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2015.; European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2015 ; Conference date: 07-09-2015 Through 11-09-2015",

year = "2015",

doi = "10.1007/978-3-319-23525-7_9",

language = "English",

isbn = "9783319235240",

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

publisher = "Springer Verlag",

pages = "137--153",

editor = "Costa, {Vitor Santos} and Carlos Soares and Annalisa Appice and Annalisa Appice and Rodrigues, {Pedro Pereira} and Costa, {Vitor Santos} and Carlos Soares and Jo{\~a}o Gama and Al{\'i}pio Jorge and Rodrigues, {Pedro Pereira} and Jo{\~a}o Gama and Costa, {Vitor Santos} and Al{\'i}pio Jorge and Annalisa Appice and Rodrigues, {Pedro Pereira} and Jo{\~a}o Gama and Annalisa Appice and Carlos Soares and Al{\'i}pio Jorge and Jo{\~a}o Gama and Rodrigues, {Pedro Pereira} and Costa, {Vitor Santos} and Carlos Soares and Al{\'i}pio Jorge",

booktitle = "Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2015",

}

TY - GEN

T1 - Opening the black box

T2 - European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2015

AU - Vidovic, Marina M.C.

AU - Görnitz, Nico

AU - Müller, Klaus Robert

AU - Rätsch, Gunnar

AU - Kloft, Marius

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2015.

PY - 2015

Y1 - 2015

N2 - This work is in the context of kernel-based learning algorithms for sequence data. We present a probabilistic approach to automatically extract, from the output of such string-kernel-based learning algorithms, the subsequences—or motifs—truly underlying the machine’s predictions. The proposed framework views motifs as free parameters in a probabilistic model, which is solved through a global optimization approach. In contrast to prevalent approaches, the proposed method can discover even difficult, long motifs, and could be combined with any kernel-based learning algorithm that is based on an adequate sequence kernel. We show that, by using a discriminate kernel machine such as a support vector machine, the approach can reveal discriminative motifs underlying the kernel predictor. We demonstrate the efficacy of our approach through a series of experiments on synthetic and real data, including problems from handwritten digit recognition and a large-scale human splice site data set from the domain of computational biology.

AB - This work is in the context of kernel-based learning algorithms for sequence data. We present a probabilistic approach to automatically extract, from the output of such string-kernel-based learning algorithms, the subsequences—or motifs—truly underlying the machine’s predictions. The proposed framework views motifs as free parameters in a probabilistic model, which is solved through a global optimization approach. In contrast to prevalent approaches, the proposed method can discover even difficult, long motifs, and could be combined with any kernel-based learning algorithm that is based on an adequate sequence kernel. We show that, by using a discriminate kernel machine such as a support vector machine, the approach can reveal discriminative motifs underlying the kernel predictor. We demonstrate the efficacy of our approach through a series of experiments on synthetic and real data, including problems from handwritten digit recognition and a large-scale human splice site data set from the domain of computational biology.

UR - http://www.scopus.com/inward/record.url?scp=84959327515&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-23525-7_9

DO - 10.1007/978-3-319-23525-7_9

M3 - Conference contribution

AN - SCOPUS:84959327515

SN - 9783319235240

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 137

EP - 153

BT - Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2015

A2 - Costa, Vitor Santos

A2 - Soares, Carlos

A2 - Appice, Annalisa

A2 - Rodrigues, Pedro Pereira

A2 - Costa, Vitor Santos

A2 - Soares, Carlos

A2 - Gama, João

A2 - Jorge, Alípio

A2 - Rodrigues, Pedro Pereira

A2 - Gama, João

A2 - Costa, Vitor Santos

A2 - Jorge, Alípio

A2 - Appice, Annalisa

A2 - Rodrigues, Pedro Pereira

A2 - Gama, João

A2 - Appice, Annalisa

A2 - Soares, Carlos

A2 - Jorge, Alípio

A2 - Gama, João

A2 - Rodrigues, Pedro Pereira

A2 - Costa, Vitor Santos

A2 - Soares, Carlos

A2 - Jorge, Alípio

PB - Springer Verlag

Y2 - 7 September 2015 through 11 September 2015

ER -

Opening the black box: Revealing interpretable sequence motifs in kernel-based learning algorithms

Abstract

Publication series

Other

Bibliographical note

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this