Berlin 2024 – wissenschaftliches Programm

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

MA: Fachverband Magnetismus

MA 46: Poster II

MA 46.4: Poster

Donnerstag, 21. März 2024, 15:00–18:00, Poster C

Gesture recognition with Brownian reservoir computing using geometrically confined skyrmion dynamics — •Grischa Beneke¹, Thomas Brian Winkler¹, Klaus Raab¹, Maarten Brems¹, Fabian Kammerbauer¹, Johan Mentink², and Mathias Kläui¹ — ¹Institut für Physik, Johannes Gutenberg-Universität Mainz, Germany — ²Radboud University, Institute for Molecules and Materials, The Netherlands

Physical Reservoir Computing (RC) represents a significant advance for reducing the training time and energy consumption in non-conventional computing. It has been shown that magnetic skyrmions, topological particle-like spin textures, are promising candidates for unconventional computing [2,3] due to their non-linear interactions and efficient mechanisms for manipulation. Leveraging the thermally activated diffusive motion [2] and utilizing an automatic reset mechanism prompted by the repulsion of skyrmions from magnetic material boundaries has facilitated the realization of spatially multiplexed RC [3]. Here we take the next step and realize time-multiplexed RC. We present a hardware-based approach to distinguish various hand gestures captured via a Doppler radar, involving the application of a time-varying voltage to observe and track the trajectory of skyrmions. Notably, our device exhibits competitive performance when compared to an energy intensive approach [4]. [1] D. Gauthier et al., Nat. Comms. 12, 5564 (2021). [2] J. Zázvorka et al., Nat. Nanotechnol. 14, 658 (2019). [3] K. Raab et al., Nat. Comms. 13, 6982 (2022). [4] G. Beneke et al., in preparation (2023).

Keywords: Skyrmion; Reservoir Computing; Gesture recognition