Die DPG-Frühjahrstagung in Dresden musste abgesagt werden! Lesen Sie mehr ...

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

HL: Fachverband Halbleiterphysik

HL 21: 2D Materials II: Electronic Structure, Excitations, etc. (joint session O/CPP/HL)

HL 21.6: Hauptvortrag

Montag, 16. März 2020, 16:15–16:45, WIL C107

Atomic scale neural circuitry capable of self-adaptation — •Brian Kiraly — Radboud University, Nijmegen, The Netherlands

Driven by the rise of artificial intelligence and its potential for reduced energy consumption, there have been expanded efforts directed toward investigating materials which can perform pattern recognition directly in hardware. This requires a step away from physical systems which show simple bistability, toward complex, stochastic systems, which are inherently tunable. At the moment, however, the state of the art in neuromorphic computing still struggles with fundamental issues, such as scaling or adaptability, often referred to as on-chip learning. In this talk, I will show a ground-breaking example, in which as few as 7 cobalt dopants on the surface of anisotropic semiconducting black phosphorus [1,2] can be utilized for pattern recognition. For this demonstration, we illustrate that controlled coupling between Co atoms [3] leads to a stochastic system which is well described using the concept of a Boltzmann machine. Both probing and reading the system with a scanning tunneling microscope, I will demonstrate how we realize an atomic scale synaptic memory and how the stochastic dynamics adapt and learn, depending on the input stimulus. The main aspects of this work have been carried out in the Scanning Probe Microscopy department at Radboud University. [1] Kiraly, Knol, Volckaert, Biswas, Rudenko, et. al., Phys. Rev. Lett. 123, 216403 (2019). [2] Kiraly, Hauptmann, Rudenko, Katsnelson, Khajetoorians, Nano Lett. 17, 3607 (2017). [3] Kiraly, Rudenko, Weerdenburg, Wegner, Katsnelson, Khajetoorians, Nature Commun. 9, 3904, (2018).