A heuristic sensitivity analysis technique for high-dimensional systems

Mustafa Berke Yelten

doi:10.1109/ICECS.2016.7841162

A heuristic sensitivity analysis technique for high-dimensional systems

Mustafa Berke Yelten^*

^*Corresponding author for this work

Department of Electronics and Communication Engineering

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

Abstract

A viable technique for sensitivity analysis in high-dimensional systems is described in the context of bio-inspired systems. The sensitivity analysis provides critical information about the system by indicating the dominant parameters that shape the output. This knowledge becomes particularly essential to have a better understanding of complex biological network of interactions. A notable feature of many high-dimensional systems is that a large portion of all parameters have little impact on the system outcome, thus yielding sparsity. The proposed algorithm leverages the sparse properties of systems analysed and is based on a heuristic two-stage elimination strategy. The implementation of the proposed algorithm yields substantial reduction of the total simulation cost by as much as 95% for a system composed of 562500 parameters over the conventional local sensitivity analysis while retaining its accuracy above 70%.

Original language	English
Title of host publication	2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	181-184
Number of pages	4
ISBN (Electronic)	9781509061136
DOIs	https://doi.org/10.1109/ICECS.2016.7841162
Publication status	Published - 2016
Event	23rd IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016 - Monte Carlo, Monaco Duration: 11 Dec 2016 → 14 Dec 2016

Publication series

Name	2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016

Conference

Conference	23rd IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016
Country/Territory	Monaco
City	Monte Carlo
Period	11/12/16 → 14/12/16

Bibliographical note

Publisher Copyright:
© 2016 IEEE.

Keywords

biological systems
high dimensional systems
Sensitivity analysis
sparse methods

Access to Document

10.1109/ICECS.2016.7841162

Cite this

Berke Yelten, M. (2016). A heuristic sensitivity analysis technique for high-dimensional systems. In 2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016 (pp. 181-184). Article 7841162 (2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICECS.2016.7841162

@inproceedings{b67a54f783d8436b8080e25c29768f46,

title = "A heuristic sensitivity analysis technique for high-dimensional systems",

abstract = "A viable technique for sensitivity analysis in high-dimensional systems is described in the context of bio-inspired systems. The sensitivity analysis provides critical information about the system by indicating the dominant parameters that shape the output. This knowledge becomes particularly essential to have a better understanding of complex biological network of interactions. A notable feature of many high-dimensional systems is that a large portion of all parameters have little impact on the system outcome, thus yielding sparsity. The proposed algorithm leverages the sparse properties of systems analysed and is based on a heuristic two-stage elimination strategy. The implementation of the proposed algorithm yields substantial reduction of the total simulation cost by as much as 95\% for a system composed of 562500 parameters over the conventional local sensitivity analysis while retaining its accuracy above 70\%.",

keywords = "biological systems, high dimensional systems, Sensitivity analysis, sparse methods",

author = "\{Berke Yelten\}, Mustafa",

note = "Publisher Copyright: {\textcopyright} 2016 IEEE.; 23rd IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016 ; Conference date: 11-12-2016 Through 14-12-2016",

year = "2016",

doi = "10.1109/ICECS.2016.7841162",

language = "English",

series = "2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "181--184",

booktitle = "2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016",

address = "United States",

}

Berke Yelten, M 2016, A heuristic sensitivity analysis technique for high-dimensional systems. in 2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016., 7841162, 2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016, Institute of Electrical and Electronics Engineers Inc., pp. 181-184, 23rd IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016, Monte Carlo, Monaco, 11/12/16. https://doi.org/10.1109/ICECS.2016.7841162

A heuristic sensitivity analysis technique for high-dimensional systems. / Berke Yelten, Mustafa.
2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016. Institute of Electrical and Electronics Engineers Inc., 2016. p. 181-184 7841162 (2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016).

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

TY - GEN

T1 - A heuristic sensitivity analysis technique for high-dimensional systems

AU - Berke Yelten, Mustafa

PY - 2016

Y1 - 2016

N2 - A viable technique for sensitivity analysis in high-dimensional systems is described in the context of bio-inspired systems. The sensitivity analysis provides critical information about the system by indicating the dominant parameters that shape the output. This knowledge becomes particularly essential to have a better understanding of complex biological network of interactions. A notable feature of many high-dimensional systems is that a large portion of all parameters have little impact on the system outcome, thus yielding sparsity. The proposed algorithm leverages the sparse properties of systems analysed and is based on a heuristic two-stage elimination strategy. The implementation of the proposed algorithm yields substantial reduction of the total simulation cost by as much as 95% for a system composed of 562500 parameters over the conventional local sensitivity analysis while retaining its accuracy above 70%.

AB - A viable technique for sensitivity analysis in high-dimensional systems is described in the context of bio-inspired systems. The sensitivity analysis provides critical information about the system by indicating the dominant parameters that shape the output. This knowledge becomes particularly essential to have a better understanding of complex biological network of interactions. A notable feature of many high-dimensional systems is that a large portion of all parameters have little impact on the system outcome, thus yielding sparsity. The proposed algorithm leverages the sparse properties of systems analysed and is based on a heuristic two-stage elimination strategy. The implementation of the proposed algorithm yields substantial reduction of the total simulation cost by as much as 95% for a system composed of 562500 parameters over the conventional local sensitivity analysis while retaining its accuracy above 70%.

KW - biological systems

KW - high dimensional systems

KW - Sensitivity analysis

KW - sparse methods

UR - https://www.scopus.com/pages/publications/85015315589

U2 - 10.1109/ICECS.2016.7841162

DO - 10.1109/ICECS.2016.7841162

M3 - Conference contribution

AN - SCOPUS:85015315589

T3 - 2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016

SP - 181

EP - 184

BT - 2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 23rd IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016

Y2 - 11 December 2016 through 14 December 2016

ER -

Berke Yelten M. A heuristic sensitivity analysis technique for high-dimensional systems. In 2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016. Institute of Electrical and Electronics Engineers Inc. 2016. p. 181-184. 7841162. (2016 IEEE International Conference on Electronics, Circuits and Systems, ICECS 2016). doi: 10.1109/ICECS.2016.7841162

A heuristic sensitivity analysis technique for high-dimensional systems

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this