| Autor: |
Rodimkov Y; Department of Mathematical Software and Supercomputing Technologies, Lobachevsky University, 603950 Nizhni Novgorod, Russia., Bhadoria S; Department of Physics, University of Gothenburg, SE-41296 Gothenburg, Sweden., Volokitin V; Department of Mathematical Software and Supercomputing Technologies, Lobachevsky University, 603950 Nizhni Novgorod, Russia.; Mathematical Center, Lobachevsky University, 603950 Nizhni Novgorod, Russia., Efimenko E; Institute of Applied Physics of the Russian Academy of Sciences, 603950 Nizhni Novgorod, Russia., Polovinkin A; Adv Learning Systems, TDATA, Intel, Chandler, AZ 85226, USA., Blackburn T; Department of Physics, University of Gothenburg, SE-41296 Gothenburg, Sweden., Marklund M; Department of Physics, University of Gothenburg, SE-41296 Gothenburg, Sweden., Gonoskov A; Department of Mathematical Software and Supercomputing Technologies, Lobachevsky University, 603950 Nizhni Novgorod, Russia.; Department of Physics, University of Gothenburg, SE-41296 Gothenburg, Sweden., Meyerov I; Department of Mathematical Software and Supercomputing Technologies, Lobachevsky University, 603950 Nizhni Novgorod, Russia.; Mathematical Center, Lobachevsky University, 603950 Nizhni Novgorod, Russia. |
| Abstrakt: |
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. |
