Original title: Distance based similarity metrics for artificial neural network estimates of soot distribution in catalytic filters
Authors: Khýr, Matyáš ; Isoz, Martin
Document type: Papers
Conference/Event: Topical Problems of Fluid Mechanics 2026, Praha (CZ), 20260218
Year: 2026
Language: eng
Abstract: Although artificial neural networks have attracted the attention of researchers as potential fast and lightweight substitutes for expensive computational models, their application to real world problems may be hindered by improper choice of training and validation metrics. In this work, we examine the applicability of so called distance based metrics to the problem of soot deposition in porous media and compare their performance with standard pixel-wise metrics such as the root mean square error. Specifically, we showcase the implementation and behavior of the Sliced Wasserstein Distance and the Sinkhorn Distance on a set of benchmarks, and we demonstrate their use with a previously developed artificial neural network for the estimation of soot deposits. Our results suggest that both the Sliced Wasserstein and Sinkhorn distances outperform the root mean squared error, however, their computational complexity may render their use in conjunction with artificial neural networks impractical.
Keywords: ANN; CF; CFD; CNN
Project no.: EH23_020/0008501, StrategieAV21/34
Funding provider: GA MŠk, AV ČR
Host item entry: Topical Problems of Fluid Mechanics 2026, ISBN 978-80-87012-92-5, ISSN 2336-5781

Institution: Institute of Thermomechanics AS ČR (web)
Document availability information: Fulltext is available at external website.
External URL: https://tpfm.it.cas.cz/fm2015/im/admin/showfile/data/my/Papers/2026/14-TPFM2026.pdf
Original record: https://hdl.handle.net/11104/0380293

Permalink: http://www.nusl.cz/ntk/nusl-781026


The record appears in these collections:
Research > Institutes ASCR > Institute of Thermomechanics
Conference materials > Papers