Towards ML-Based Diagnostics of Laser–Plasma Interactions
The power of machine learning (ML) in feature identification can be harnessed for determining quantities in experiments that are difficult to measure directly. However, if an ML model is trained on simulated data, rather than experimental results, the differences between the two can pose an obstacle...
Main Authors: | , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2021-10-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/21/21/6982 |
_version_ | 1797511768029790208 |
---|---|
author | Yury Rodimkov Shikha Bhadoria Valentin Volokitin Evgeny Efimenko Alexey Polovinkin Thomas Blackburn Mattias Marklund Arkady Gonoskov Iosif Meyerov |
author_facet | Yury Rodimkov Shikha Bhadoria Valentin Volokitin Evgeny Efimenko Alexey Polovinkin Thomas Blackburn Mattias Marklund Arkady Gonoskov Iosif Meyerov |
author_sort | Yury Rodimkov |
collection | DOAJ |
description | The power of machine learning (ML) in feature identification can be harnessed for determining quantities in experiments that are difficult to measure directly. However, if an ML model is trained on simulated data, rather than experimental results, the differences between the two can pose an obstacle to reliable data extraction. Here we report on the development of ML-based diagnostics for experiments on high-intensity laser–matter interactions. With the intention to accentuate robust, physics-governed features, the presence of which is tolerant to such differences, we test the application of principal component analysis, data augmentation and training with data that has superimposed noise of gradually increasing amplitude. Using synthetic data of simulated experiments, we identify that the approach based on the noise of increasing amplitude yields the most accurate ML models and thus is likely to be useful in similar projects on ML-based diagnostics. |
first_indexed | 2024-03-10T05:52:43Z |
format | Article |
id | doaj.art-3c3b37f45d0047e7bd7ebe672cf0bdbb |
institution | Directory Open Access Journal |
issn | 1424-8220 |
language | English |
last_indexed | 2024-03-10T05:52:43Z |
publishDate | 2021-10-01 |
publisher | MDPI AG |
record_format | Article |
series | Sensors |
spelling | doaj.art-3c3b37f45d0047e7bd7ebe672cf0bdbb2023-11-22T21:34:49ZengMDPI AGSensors1424-82202021-10-012121698210.3390/s21216982Towards ML-Based Diagnostics of Laser–Plasma InteractionsYury Rodimkov0Shikha Bhadoria1Valentin Volokitin2Evgeny Efimenko3Alexey Polovinkin4Thomas Blackburn5Mattias Marklund6Arkady Gonoskov7Iosif Meyerov8Department of Mathematical Software and Supercomputing Technologies, Lobachevsky University, 603950 Nizhni Novgorod, RussiaDepartment of Physics, University of Gothenburg, SE-41296 Gothenburg, SwedenDepartment of Mathematical Software and Supercomputing Technologies, Lobachevsky University, 603950 Nizhni Novgorod, RussiaInstitute of Applied Physics of the Russian Academy of Sciences, 603950 Nizhni Novgorod, RussiaAdv Learning Systems, TDATA, Intel, Chandler, AZ 85226, USADepartment of Physics, University of Gothenburg, SE-41296 Gothenburg, SwedenDepartment of Physics, University of Gothenburg, SE-41296 Gothenburg, SwedenDepartment of Mathematical Software and Supercomputing Technologies, Lobachevsky University, 603950 Nizhni Novgorod, RussiaDepartment of Mathematical Software and Supercomputing Technologies, Lobachevsky University, 603950 Nizhni Novgorod, RussiaThe power of machine learning (ML) in feature identification can be harnessed for determining quantities in experiments that are difficult to measure directly. However, if an ML model is trained on simulated data, rather than experimental results, the differences between the two can pose an obstacle to reliable data extraction. Here we report on the development of ML-based diagnostics for experiments on high-intensity laser–matter interactions. With the intention to accentuate robust, physics-governed features, the presence of which is tolerant to such differences, we test the application of principal component analysis, data augmentation and training with data that has superimposed noise of gradually increasing amplitude. Using synthetic data of simulated experiments, we identify that the approach based on the noise of increasing amplitude yields the most accurate ML models and thus is likely to be useful in similar projects on ML-based diagnostics.https://www.mdpi.com/1424-8220/21/21/6982laser–plasmamachine learningneural networkdimension reductiondata augmentation |
spellingShingle | Yury Rodimkov Shikha Bhadoria Valentin Volokitin Evgeny Efimenko Alexey Polovinkin Thomas Blackburn Mattias Marklund Arkady Gonoskov Iosif Meyerov Towards ML-Based Diagnostics of Laser–Plasma Interactions Sensors laser–plasma machine learning neural network dimension reduction data augmentation |
title | Towards ML-Based Diagnostics of Laser–Plasma Interactions |
title_full | Towards ML-Based Diagnostics of Laser–Plasma Interactions |
title_fullStr | Towards ML-Based Diagnostics of Laser–Plasma Interactions |
title_full_unstemmed | Towards ML-Based Diagnostics of Laser–Plasma Interactions |
title_short | Towards ML-Based Diagnostics of Laser–Plasma Interactions |
title_sort | towards ml based diagnostics of laser plasma interactions |
topic | laser–plasma machine learning neural network dimension reduction data augmentation |
url | https://www.mdpi.com/1424-8220/21/21/6982 |
work_keys_str_mv | AT yuryrodimkov towardsmlbaseddiagnosticsoflaserplasmainteractions AT shikhabhadoria towardsmlbaseddiagnosticsoflaserplasmainteractions AT valentinvolokitin towardsmlbaseddiagnosticsoflaserplasmainteractions AT evgenyefimenko towardsmlbaseddiagnosticsoflaserplasmainteractions AT alexeypolovinkin towardsmlbaseddiagnosticsoflaserplasmainteractions AT thomasblackburn towardsmlbaseddiagnosticsoflaserplasmainteractions AT mattiasmarklund towardsmlbaseddiagnosticsoflaserplasmainteractions AT arkadygonoskov towardsmlbaseddiagnosticsoflaserplasmainteractions AT iosifmeyerov towardsmlbaseddiagnosticsoflaserplasmainteractions |