ARC@ORU: Perspectives on and beyond Reservoir Computing

ARC@ORU Research Seminar Series

Perspectives on and beyond Reservoir Computing

This is a Key Note seminar within the kick-off workshop for the Unconventional Computing Nordic Research Network, UniCoRN.

About the talk

This talk revisits reservoir computing as an alternative paradigm for sequence modeling, motivated by the well-known difficulties of training recurrent neural networks and the growing demand for efficient, hardware-compatible learning systems. After introducing the core principles of reservoir computing and its natural suitability for neuromorphic and physical implementations, the limitations of conventional echo state networks in terms of long-range information propagation are discussed. The second part of the talk presents recent research directions focusing on non-dissipative reservoir dynamics and novel reservoir architectures that extend reservoir computing toward the deep learning regime while preserving fast training and hardware amenability.

Speaker

Claudio Gallicchio is an Associate Professor of Computer Science at the University of Pisa. His research focuses on reservoir computing, recurrent neural networks, and dynamical systems for deep learning, with particular emphasis on hardware-aware and energy-efficient models. He is involved in several research projects on merging concepts from deep learning and neuromorphic hardware for sustainable artificial intelligence.