Tipo evento: Seminario INFN

Data e ora: Mercoledi’ 15/11/2023 ore 16.30

Relatore: Alessandro Ingrosso (ICTP, Trieste)

Titolo: A computation-dissipation tradeoff for machine learning at the mesoscale

Dove: Aula Maxwell, Plesso di Fisica, Dipartimento di Scienze Matematiche, Fisiche e Informatiche

 

Abstract: The cost of information processing in physical systems calls for a trade-off between performance and energetic expenditure. We formulate a computation-dissipation bottleneck in mesoscopic systems used as input-output devices. Using both real datasets and synthetic tasks, we show how non-equilibrium leads to enhanced performance. Our framework sheds light on a crucial compromise between information compression, input-output computation and dynamic irreversibility induced by non-reciprocal interactions.

Email organizzatori: pietro.rotondo@unipr.it  raffaella.burioni@unipr.it

Modified on