Berlin 2024 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
DY: Fachverband Dynamik und Statistische Physik
DY 49: Focus Session: Computing with Dynamical Systems: New Perspectives on Reservoirs and Applications II – Applications and Quantum RC
DY 49.3: Talk
Thursday, March 21, 2024, 15:45–16:00, BH-N 243
Memristive devices for chaotic time series prediction — •Kristina Nikiruy1, Seongae Park1, Tzvetan Ivanov1,2, Alon Ascoli3, Fernando Corinto4, Ronald Tetzlaff3, and Martin Ziegler1,2 — 1Micro- and Nanoelectronic Systems, Department of Electrical Engineering and Information Technology, TU Ilmenau, Germany — 2Institute of Micro- and Nanotechnologies MacroNano*, TU Ilmenau, Germany — 3Faculty of Electrical and Computer Engineering, Institute of Circuits and Systems, TU Dresden, Germany — 4Department of Electronics and Telecommunications (DET), Politecnico di Torino, Turin, Italy
Neuromorphic hardware, in which memristive devices are key elements, enables an energy-efficient and compact realization of the signal processing concept. Here we show that the application of nonlinear vector autoregression (NVAR), also known as next generation reservoir computing (NGRC), to a single-layer network with memristive weights can be used to predict signals, depending on the nature of the nonlinear functions and the number of weights. The network has been experimentally implemented with HfO2-based memristive devices and allows an accurate prediction of chaotic time series of Mackey-Glass and Duffing oscillators. The effect of the nonlinear combinations of the input data points, the network structure, and the number and nonlinear resistive switching properties of the memristive weights on the output response are studied. In this regard, it is shown how a suitable network structure can be tailored for chaotic time series prediction.
Keywords: memristive devices; chaotic time series; autoregression