pages_top_content_img_9.jpg
Logo_JS_207.jpg

Réseaux du futur : 5G et au-delà

11-13 mars, 2020
Telecom Paris, Institut Polytechnique de Paris, Palaiseau 

Jeudi 12
Exposition et dosimétrie dans le contexte de la 5G

› 14:00 - 15:20 (1h20)
› Forum / Posters
Artificial Neural Networks for Uncertainty Quantification in RF Radiation Modeling
Xi Cheng  1@  , Clément Henry  2@  , Francesco Andriulli  2@  , Christian Person  3@  , Joe Wiart  1@  
1 : Chaire C2M  (C2M)
LTCI, Télécom ParisTech
2 : Politecnico di Torino  (Politecnico di Torino)
3 : IMT Atlantique / lab-STICC UMR CNRS 6285  (lab-STICC UMR)
IMT Atlantique

This paper focuses on quantifying the uncertainty in the outputs of numerical simulations produced by uncertain positions of the electrodes placed on patient's scalp. In order to avoid running thousands of simulations, an architecture which combines two different artificial neural networks (ANNs) for uncertainty quantification (UQ) is proposed in this paper. The overall aim of this work is to develop a surrogate model for UQ involving high-dimensional data. The proposed method is demonstrated to be an attractive alternative to conventional UQ methods since it shows considerable advantage in the computational expense and speed.


Personnes connectées : 1