Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

DY: Fachverband Dynamik und Statistische Physik

DY 42: Focus Session: Computing with Dynamical Systems: New Perspectives on Reservoirs and Applications I – Fundamentals

DY 42.8: Vortrag

Donnerstag, 21. März 2024, 11:45–12:00, BH-N 243

Designing Active Matter Systems for Reservoir Computing — •Mario U. Gaimann and Miriam Klopotek — Stuttgart Center for Simulation Science (SimTech), Cluster of Excellence EXC 2075, University of Stuttgart, Germany

Reservoir computing with physical systems is a candidate for next-generation computing [1,2]. It is a powerful method to solve challenging tasks such as chaotic time-series prediction. However, tuning an arbitrary reservoir – which may be physical – for optimal properties like its dynamical regime remains an open question. To approach this problem we use a novel flavour of reservoir commuting based on simple active matter models [3]. We systematically study how the predictive performance of our driven active matter reservoir depends on a variety of physical hyper-parameters – the number of agents, the extent of driver-reservoir interaction, as well as different noise types and forces. For each set, we characterize the spatio-temporal, heterogeneous, yet also collective dynamics of the swarm. We aim to understand optimal conditions for learning and inspire new forms of physical, natural, and bio-inspired computing.
 [1] Tanaka, G. et al. (2019), Neural Networks 115, 100-123.
 [2] Nakajima, K. and Fischer, I. (2021). Reservoir Computing. Springer Singapore.
 [3] Lymburn, T. et al. (2021), Chaos 31(3), 033121.

Keywords: Reservoir Computing; Active Matter; Non-equilibrium Dynamics; Time-Series Prediction; Machine Learning