Trustworthiness Score for Echo State Networks by Analysis of the Reservoir Dynamics

  1. Enguita-Gonzalez, Jose M. 1
  2. Garcia-Perez, Diego 1
  3. Cuadrado-Vega, Abel Alberto 1
  4. García-Peña, Daniel
  5. Rodríguez-Ossorio, José Ramón
  6. Diaz-Blanco, Ignacio 1
  1. 1 Universidad de Oviedo
    info

    Universidad de Oviedo

    Oviedo, España

    ROR https://ror.org/006gksa02

Actas:
European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN 2024

Año de publicación: 2024

Páginas: 455-460

Congreso: European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (32th. 2024. Bruges, Bélgica)

Tipo: Aportación congreso

DOI: 10.14428/ESANN/2024.ES2024-38 GOOGLE SCHOLAR lock_openAcceso abierto editor

Resumen

Epistemic uncertainty arises from input data areas where models lack exposure during training and may result in significant performance degradation in deployment. Echo State Networks are often used as virtual sensors or digital twins processing temporal input data, so their robustness against this degradation is crucial. This paper addresses this challenge by proposing a score comparing the similarity between the dynamic evolution of the reservoir in training and in inference. This research aims to enhance model confidence and adaptability in evolving circumstances.

€ Ver financiación

Información de financiación

This work is part of Grant PID2020-115401GB-I00 funded by MCIN/AEI/ 10.13039/501100011033.

Financiadores