Towards explaining anomalies: A deep Taylor decomposition of one-class models

Jacob Kauffmann; Klaus Robert Müller; Grégoire Montavon

doi:10.1016/j.patcog.2020.107198

Towards explaining anomalies: A deep Taylor decomposition of one-class models

Jacob Kauffmann, Klaus Robert Müller, Grégoire Montavon

Research output: Contribution to journal › Article › peer-review

64 Citations (Scopus)

Abstract

Detecting anomalies in the data is a common machine learning task, with numerous applications in the sciences and industry. In practice, it is not always sufficient to reach high detection accuracy, one would also like to be able to understand why a given data point has been predicted to be anomalous. We propose a principled approach for one-class SVMs (OC-SVM), that draws on the novel insight that these models can be rewritten as distance/pooling neural networks. This ‘neuralization’ step lets us apply deep Taylor decomposition (DTD), a methodology that leverages the model structure in order to quickly and reliably explain decisions in terms of input features. The proposed method (called ‘OC-DTD’) is applicable to a number of common distance-based kernel functions, and it outperforms baselines such as sensitivity analysis, distance to nearest neighbor, or edge detection.

Original language	English
Article number	107198
Journal	Pattern Recognition
Volume	101
DOIs	https://doi.org/10.1016/j.patcog.2020.107198
Publication status	Published - 2020 May

Bibliographical note

Funding Information:
This research was supported by the Institute for Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (No. 2017-0-00451, No. 2017-0-01779); the Deutsche Forschungsgemeinschaft (DFG) [grant MU 987/17-1]; the German Ministry for Education and Research as Berlin Big Data Center (BBDC) [01IS14013A] and Berlin Center for Machine Learning (BZML) [01IS18037A]. We are grateful to Guido Schwenk for the valuable discussion.

Funding Information:
This research was supported by the Institute for Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (No. 2017-0-00451 , No. 2017-0-01779 ); the Deutsche Forschungsgemeinschaft (DFG) [grant MU 987/17-1 ]; the German Ministry for Education and Research as Berlin Big Data Center (BBDC) [01IS14013A] and Berlin Center for Machine Learning (BZML) [01IS18037A]. We are grateful to Guido Schwenk for the valuable discussion.

Publisher Copyright:
© 2020

Keywords

Deep Taylor decomposition
Explainable machine learning
Kernel machines
Outlier detection
Unsupervised learning

ASJC Scopus subject areas

Software
Signal Processing
Computer Vision and Pattern Recognition
Artificial Intelligence

Access to Document

10.1016/j.patcog.2020.107198

Cite this

@article{a1e2d62161b640718eee48eb8404122a,

title = "Towards explaining anomalies: A deep Taylor decomposition of one-class models",

abstract = "Detecting anomalies in the data is a common machine learning task, with numerous applications in the sciences and industry. In practice, it is not always sufficient to reach high detection accuracy, one would also like to be able to understand why a given data point has been predicted to be anomalous. We propose a principled approach for one-class SVMs (OC-SVM), that draws on the novel insight that these models can be rewritten as distance/pooling neural networks. This {\textquoteleft}neuralization{\textquoteright} step lets us apply deep Taylor decomposition (DTD), a methodology that leverages the model structure in order to quickly and reliably explain decisions in terms of input features. The proposed method (called {\textquoteleft}OC-DTD{\textquoteright}) is applicable to a number of common distance-based kernel functions, and it outperforms baselines such as sensitivity analysis, distance to nearest neighbor, or edge detection.",

keywords = "Deep Taylor decomposition, Explainable machine learning, Kernel machines, Outlier detection, Unsupervised learning",

author = "Jacob Kauffmann and M{\"u}ller, {Klaus Robert} and Gr{\'e}goire Montavon",

note = "Funding Information: This research was supported by the Institute for Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (No. 2017-0-00451, No. 2017-0-01779); the Deutsche Forschungsgemeinschaft (DFG) [grant MU 987/17-1]; the German Ministry for Education and Research as Berlin Big Data Center (BBDC) [01IS14013A] and Berlin Center for Machine Learning (BZML) [01IS18037A]. We are grateful to Guido Schwenk for the valuable discussion. Funding Information: This research was supported by the Institute for Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (No. 2017-0-00451 , No. 2017-0-01779 ); the Deutsche Forschungsgemeinschaft (DFG) [grant MU 987/17-1 ]; the German Ministry for Education and Research as Berlin Big Data Center (BBDC) [01IS14013A] and Berlin Center for Machine Learning (BZML) [01IS18037A]. We are grateful to Guido Schwenk for the valuable discussion. Publisher Copyright: {\textcopyright} 2020",

year = "2020",

month = may,

doi = "10.1016/j.patcog.2020.107198",

language = "English",

volume = "101",

journal = "Pattern Recognition",

issn = "0031-3203",

publisher = "Elsevier Ltd",

}

TY - JOUR

T1 - Towards explaining anomalies

T2 - A deep Taylor decomposition of one-class models

AU - Kauffmann, Jacob

AU - Müller, Klaus Robert

AU - Montavon, Grégoire

N1 - Funding Information: This research was supported by the Institute for Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (No. 2017-0-00451, No. 2017-0-01779); the Deutsche Forschungsgemeinschaft (DFG) [grant MU 987/17-1]; the German Ministry for Education and Research as Berlin Big Data Center (BBDC) [01IS14013A] and Berlin Center for Machine Learning (BZML) [01IS18037A]. We are grateful to Guido Schwenk for the valuable discussion. Funding Information: This research was supported by the Institute for Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (No. 2017-0-00451 , No. 2017-0-01779 ); the Deutsche Forschungsgemeinschaft (DFG) [grant MU 987/17-1 ]; the German Ministry for Education and Research as Berlin Big Data Center (BBDC) [01IS14013A] and Berlin Center for Machine Learning (BZML) [01IS18037A]. We are grateful to Guido Schwenk for the valuable discussion. Publisher Copyright: © 2020

PY - 2020/5

Y1 - 2020/5

N2 - Detecting anomalies in the data is a common machine learning task, with numerous applications in the sciences and industry. In practice, it is not always sufficient to reach high detection accuracy, one would also like to be able to understand why a given data point has been predicted to be anomalous. We propose a principled approach for one-class SVMs (OC-SVM), that draws on the novel insight that these models can be rewritten as distance/pooling neural networks. This ‘neuralization’ step lets us apply deep Taylor decomposition (DTD), a methodology that leverages the model structure in order to quickly and reliably explain decisions in terms of input features. The proposed method (called ‘OC-DTD’) is applicable to a number of common distance-based kernel functions, and it outperforms baselines such as sensitivity analysis, distance to nearest neighbor, or edge detection.

AB - Detecting anomalies in the data is a common machine learning task, with numerous applications in the sciences and industry. In practice, it is not always sufficient to reach high detection accuracy, one would also like to be able to understand why a given data point has been predicted to be anomalous. We propose a principled approach for one-class SVMs (OC-SVM), that draws on the novel insight that these models can be rewritten as distance/pooling neural networks. This ‘neuralization’ step lets us apply deep Taylor decomposition (DTD), a methodology that leverages the model structure in order to quickly and reliably explain decisions in terms of input features. The proposed method (called ‘OC-DTD’) is applicable to a number of common distance-based kernel functions, and it outperforms baselines such as sensitivity analysis, distance to nearest neighbor, or edge detection.

KW - Deep Taylor decomposition

KW - Explainable machine learning

KW - Kernel machines

KW - Outlier detection

KW - Unsupervised learning

UR - http://www.scopus.com/inward/record.url?scp=85078086602&partnerID=8YFLogxK

U2 - 10.1016/j.patcog.2020.107198

DO - 10.1016/j.patcog.2020.107198

M3 - Article

AN - SCOPUS:85078086602

SN - 0031-3203

VL - 101

JO - Pattern Recognition

JF - Pattern Recognition

M1 - 107198

ER -

Towards explaining anomalies: A deep Taylor decomposition of one-class models

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this