Learning with analytical models

Huda Ibeid, Siping Meng, Oliver Dobon, Luke Olson, William D Gropp

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

To understand and predict the performance of scientific applications, several analytical and machine learning approaches have been proposed, each having its advantages and disadvantages. In this paper, we propose and validate a hybrid approach for performance modeling and prediction, which combines analytical and machine learning models. The proposed hybrid model aims to minimize prediction cost while providing reasonable prediction accuracy. Our validation results show that the hybrid model is able to learn and correct the analytical models to better match the actual performance. Furthermore, the proposed hybrid model improves the prediction accuracy in comparison to pure machine learning techniques while using small training datasets, thus making it suitable for hardware and workload changes.

Original languageEnglish (US)
Title of host publicationProceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages778-786
Number of pages9
ISBN (Electronic)9781728135106
DOIs
StatePublished - May 2019
Event33rd IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019 - Rio de Janeiro, Brazil
Duration: May 20 2019May 24 2019

Publication series

NameProceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019

Conference

Conference33rd IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019
CountryBrazil
CityRio de Janeiro
Period5/20/195/24/19

Fingerprint

Hybrid Model
Analytical Model
Analytical models
Machine Learning
Learning systems
Prediction
Performance Modeling
Performance Prediction
Hybrid Approach
Workload
Hardware
Minimise
Predict
Costs
Learning
Analytical model
Hybrid model
Machine learning
Prediction accuracy
Model

Keywords

  • Analytical modeling
  • Hybrid modeling
  • Machine learning
  • Performance prediction

ASJC Scopus subject areas

  • Information Systems and Management
  • Artificial Intelligence
  • Computer Networks and Communications
  • Hardware and Architecture
  • Control and Optimization

Cite this

Ibeid, H., Meng, S., Dobon, O., Olson, L., & Gropp, W. D. (2019). Learning with analytical models. In Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019 (pp. 778-786). [8778229] (Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IPDPSW.2019.00128

Learning with analytical models. / Ibeid, Huda; Meng, Siping; Dobon, Oliver; Olson, Luke; Gropp, William D.

Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019. Institute of Electrical and Electronics Engineers Inc., 2019. p. 778-786 8778229 (Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Ibeid, H, Meng, S, Dobon, O, Olson, L & Gropp, WD 2019, Learning with analytical models. in Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019., 8778229, Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019, Institute of Electrical and Electronics Engineers Inc., pp. 778-786, 33rd IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019, Rio de Janeiro, Brazil, 5/20/19. https://doi.org/10.1109/IPDPSW.2019.00128
Ibeid H, Meng S, Dobon O, Olson L, Gropp WD. Learning with analytical models. In Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019. Institute of Electrical and Electronics Engineers Inc. 2019. p. 778-786. 8778229. (Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019). https://doi.org/10.1109/IPDPSW.2019.00128
Ibeid, Huda ; Meng, Siping ; Dobon, Oliver ; Olson, Luke ; Gropp, William D. / Learning with analytical models. Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 778-786 (Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019).
@inproceedings{8d4a8dd0420c4dd890af668159ef9330,
title = "Learning with analytical models",
abstract = "To understand and predict the performance of scientific applications, several analytical and machine learning approaches have been proposed, each having its advantages and disadvantages. In this paper, we propose and validate a hybrid approach for performance modeling and prediction, which combines analytical and machine learning models. The proposed hybrid model aims to minimize prediction cost while providing reasonable prediction accuracy. Our validation results show that the hybrid model is able to learn and correct the analytical models to better match the actual performance. Furthermore, the proposed hybrid model improves the prediction accuracy in comparison to pure machine learning techniques while using small training datasets, thus making it suitable for hardware and workload changes.",
keywords = "Analytical modeling, Hybrid modeling, Machine learning, Performance prediction",
author = "Huda Ibeid and Siping Meng and Oliver Dobon and Luke Olson and Gropp, {William D}",
year = "2019",
month = "5",
doi = "10.1109/IPDPSW.2019.00128",
language = "English (US)",
series = "Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "778--786",
booktitle = "Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019",
address = "United States",

}

TY - GEN

T1 - Learning with analytical models

AU - Ibeid, Huda

AU - Meng, Siping

AU - Dobon, Oliver

AU - Olson, Luke

AU - Gropp, William D

PY - 2019/5

Y1 - 2019/5

N2 - To understand and predict the performance of scientific applications, several analytical and machine learning approaches have been proposed, each having its advantages and disadvantages. In this paper, we propose and validate a hybrid approach for performance modeling and prediction, which combines analytical and machine learning models. The proposed hybrid model aims to minimize prediction cost while providing reasonable prediction accuracy. Our validation results show that the hybrid model is able to learn and correct the analytical models to better match the actual performance. Furthermore, the proposed hybrid model improves the prediction accuracy in comparison to pure machine learning techniques while using small training datasets, thus making it suitable for hardware and workload changes.

AB - To understand and predict the performance of scientific applications, several analytical and machine learning approaches have been proposed, each having its advantages and disadvantages. In this paper, we propose and validate a hybrid approach for performance modeling and prediction, which combines analytical and machine learning models. The proposed hybrid model aims to minimize prediction cost while providing reasonable prediction accuracy. Our validation results show that the hybrid model is able to learn and correct the analytical models to better match the actual performance. Furthermore, the proposed hybrid model improves the prediction accuracy in comparison to pure machine learning techniques while using small training datasets, thus making it suitable for hardware and workload changes.

KW - Analytical modeling

KW - Hybrid modeling

KW - Machine learning

KW - Performance prediction

UR - http://www.scopus.com/inward/record.url?scp=85070403222&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85070403222&partnerID=8YFLogxK

U2 - 10.1109/IPDPSW.2019.00128

DO - 10.1109/IPDPSW.2019.00128

M3 - Conference contribution

AN - SCOPUS:85070403222

T3 - Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019

SP - 778

EP - 786

BT - Proceedings - 2019 IEEE 33rd International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2019

PB - Institute of Electrical and Electronics Engineers Inc.

ER -