A hybrid objective function for robustness of artificial neural networks-estimation of parameters in a mechanical system. ETNA - Electronic Transactions on Numerical Analysis
Fiche du document
14 mars 2022
- handle: 10670/1.smuxjr
- issn: 1068-9613
- doi: 10.1553/etna_vol56s209
info:eu-repo/semantics/openAccess
Sujets proches
Identification, System Science and space Space research Natural philosophy Philosophy, Natural Exploration of space Space research Solar system--Exploration Space exploration (Astronautics) Modelmaking Modelling Model-making Molding (Clay, plaster, etc.) Clay modeling Modelling Analysis (Mathematics) Prediction Forecasts Futurology Family Family structure Families--Social aspects Families--Social conditions Family life Structure, Family Family relationships Relationships, Family Pattern Model Kinetics Dynamical systems Western architecture (Western countries) Building design Construction Buildings--Design and construction Architecture, Western (Western countries) Biography--Ancestry Relations with family Biography--Descendants Biography--Family AncestryCiter ce document
Hans-Peter Beise et al., « A hybrid objective function for robustness of artificial neural networks-estimation of parameters in a mechanical system. ETNA - Electronic Transactions on Numerical Analysis », Elektronisches Publikationsportal der Österreichischen Akademie der Wissenschafte, ID : 10.1553/etna_vol56s209
Métriques
Partage / Export
Résumé
In several studies, hybrid neural networks have proven to be more robust against noisy input data compared to plain data driven neural networks. We consider the task of estimating parameters of a mechanical vehicle model based on acceleration profiles. We introduce a convolutional neural network architecture that given sequential data, is capable to predict the parameters for a family of vehicle models that differ in the unknown parameters. This network is trained with two objective functions. The first one constitutes a more naive approach that assumes that the true parameters are known. The second objective incorporates the knowledge of the underlying dynamics and is therefore considered as hybrid approach. We show that in terms of robustness, the latter outperforms the first objective on unknown noisy input data.