Upcoming events

NBPS: Dr. Lukas Gonon

Location: JCMB Lecture Theater C

Description: Title: Reservoir Computing with Stochastic Inputs: Universality, Error Bounds and Financial Applications

Abstract: We study dynamic machine learning for discrete-time stochastic processes based on reservoir computing. Putting particular emphasis on echo state networks, we present results on universal approximation properties as well as error bounds for learning tasks based on these systems. Finally, we apply them to the problem of predicting realized covariances of financial time series. The talk is based on joint works with Juan-Pablo Ortega and Lyudmila Grigoryeva.