Recent News
Mechanical Engineering to host seminar on reservoir computing
February 28, 2025
From fireflies to drones: UNM researchers uncover strategy for synchronization efficiency
October 30, 2024
Engineering alumnus joins McLaren Formula 1 team
September 13, 2024
Two Engineering undergrads receive New Mexico Space Grant Consortium Scholarships
July 8, 2024
News Archives
Mechanical Engineering to host seminar on reservoir computing
February 28, 2025 - By Carly Bowling
The University of New Mexico Department of Mechanical Engineering will host a seminar delivered by Louis Pecora, a research scientist at the Institute for Research in Electronics & Applied Physics at University of Maryland titled, “Statistics and Dynamics of Attractor Embeddings in Reservoir Computing.” The lecture is scheduled to take place Friday, March 7 at 3:30 p.m. in Centennial Engineering Center’s Larrañaga Auditorium.
Reservoir computing is a recent branch of artificial intelligence and neural network research that has emerged as a new direction for signal analysis. When driven by a single time series from complex systems, reservoir computers can recreate all time-dependent signals of the system. Researchers have speculated that the reservoir computers may be a fundamental type of system that describes how neuronal networks in biology process sensory input. Mathematical advances have moved this argument closer to general theorem, but reservoir computers constructed from physical systems like lasers, analog circuits and possibly neural networks are not well-understood at all.
In his seminar, Pecora will discuss research processes utilized to better understand reservoir computers, including two statistics developed for other uses that can help test for homeomorphisms and diffeomorphisms between a drive system and the reservoir computer by using the time series from both systems. The viewpoints and results described by Pecora help clarify concepts about reservoir computers that are only vaguely understood, like fading memory, stability and types of dynamics that are useful.
For more information, please contact Professor Francesco Sorrentino at fsorrent@unm.edu.