Contact person: prof. dr. Herbert Jaeger
Methods from machine learning, neural networks, and dynamical systems theory are combined to develop new classes of computational processes that can operate on non-digital "brain-like" ("neuromorphic") hardware. Such neuro-inspired microchip technologies are a core theme in the CogniGron center. Research in this area is necessarily interdisciplinary, combining efforts from the computing sciences, AI, mathematics, physics and materials science. The contributions from APS focus on mathematical models of computing in non-digital "brain-like" microchips, on learning algorithms that can operate on analog spiking neural network hardware, and on homeostatic self-regulation of such systems to make them robust against temperature changes, low numerical precision, and physical aging.
|Last modified:||24 November 2021 11.23 a.m.|