BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Machine learning in chemistry and beyond" (ChE-605) seminar by Mat
 hieu Baltussen: Information processing with chemical reservoir computers
DTSTART:20251014T151500
DTEND:20251014T160000
DTSTAMP:20260921T204405Z
UID:3ac569436339af65b739f16ebf16c0897df0d040f2759c9571b66bac
CATEGORIES:Conferences - Seminars
DESCRIPTION:Mathieu Baltussen recently completed his PhD at the Institut
 e for Molecules and Materials\, Radboud University in the Netherlands. His
  research focused on information processing in chemical networks\, resulti
 ng in the first reported chemical-based and protein-based reservoir comput
 ers. He also developed data-driven learning and optimization algorithms fo
 r synthetic enzyme pathways. Before his PhD he obtained cum laude double M
 Sc degrees in physics and chemistry at Utrecht University and spent a year
  as an Erasmus research student investigating grain boundary dynamics in l
 iquid crystals in the Colloid group at Oxford University.      \nChemic
 al reaction networks\, such as those found in metabolism and signalling pa
 thways\, enable cells to process information from their environment and pe
 rform computations. Recently\, we reported on the creation of self-organiz
 ing chemical reaction networks that share similar computational capabiliti
 es. Using a physical reservoir computing approach\, we demonstrate how the
 se complex chemical systems can perform several nonlinear classification t
 asks in parallel\, predict the dynamics of other complex systems and achie
 ve time-series forecasting. We further evaluate the reservoir’s short-te
 rm memory by quantifying the predictive information of every compound in t
 he network. Our work provides proof of principle for the emergent computat
 ional capabilities of complex chemical reaction networks\, paving the way 
 for a new class of biomimetic\, intelligent materials and alternative form
 s of computational hardware.
LOCATION:https://epfl.zoom.us/j/68447908297?pwd=OU5JUGJUSUhZc0ZNYjQ2WENvYI
 NRdz09
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR