Dynamically writing coupled memories using a reinforcement learning agent, meeting physical bounds - CNRS - Centre national de la recherche scientifique Access content directly
Journal Articles Communications Physics Year : 2022

Dynamically writing coupled memories using a reinforcement learning agent, meeting physical bounds

Laura Michel
Adèle Douin

Abstract

Bits manipulation in traditional memory writing is commonly done through quasi-static operations. While simple to model, this method is known to reduce memory capacity. We demonstrate how a reinforcement learning agent can exploit the dynamical response of a simple multi-bit mechanical system to restore its memory. To do so, we introduce a model framework consisting of a chain of bi-stable springs manipulated on one end by the external action of the agent. We show that the agent learns how to reach all available states for three springs, even though some states are not reachable through adiabatic manipulation, and that training is significantly improved using transfer learning techniques. Interestingly, the agent also points to an optimal system design by taking advantage of the underlying physics. Indeed, the control time exhibits a non-monotonic dependence on the internal dissipation, reaching a minimum at a cross-over shown to verify a mechanically motivated scaling relation.
Fichier principal
Vignette du fichier
Draft - Clean.pdf (1.19 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-04234809 , version 1 (10-10-2023)

Licence

Identifiers

Cite

Théo Jules, Laura Michel, Adèle Douin, Frédéric Lechenault. Dynamically writing coupled memories using a reinforcement learning agent, meeting physical bounds. Communications Physics, 2022, 6 (1), pp.25. ⟨10.1038/s42005-023-01142-y⟩. ⟨hal-04234809⟩
11 View
10 Download

Altmetric

Share

Gmail Mastodon Facebook X LinkedIn More