Dresden 2026 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

SYBT: Symposium Beyond Transistors: Material-Based Edge Computing Paradigms

SYBT 1: Beyond Transistors: Material-Based Edge Computing Paradigms

SYBT 1.4: Invited Talk

Wednesday, March 11, 2026, 11:15–11:45, HSZ/AUDI

Tackling Reliability and Scalability in Neuromorphic Computing via Noise-aware Learning — •Eleni Vasilaki — Computer Science, The University of Sheffield, 211 Portobello, Sheffield, S1 4DP, UK

Neuromorphic computing has evolved into a broad label covering technologies ranging from biologically inspired systems to machine-learning-style architectures. Computing with materials is often promoted as a potentially energy-efficient alternative to conventional hardware, but some of these claims lack robust empirical support. In this talk I will outline key challenges in the field, with a focus on variability and scalability. I will present concrete examples from my recent work showing how noise-aware learning in systems with stochastic, device-level behaviour can help mitigate variability and improve robustness. These results suggest that while variability poses real constraints, it can be addressed through appropriate learning strategies.

Keywords: Neural SDEs; Noise Aware modelling