Regensburg 2022 – scientific programme

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

DY: Fachverband Dynamik und Statistische Physik

DY 45: Poster Session: Nonlinear Dynamics, Pattern Formation, Data Analytics and Machine Learning

DY 45.10: Poster

Thursday, September 8, 2022, 15:00–18:00, P2

Influence of delay-times on photonics reservoir computing performance — •Lina Jaurigue and Kathy Lüdge — Institut f. Physik, Technische Universität Ilmenau, Weimarer Str. 25, 98684 Ilmenau, Germany

Reservoir computing is a machine learning approach that utilises the non-linear responses of dynamical systems to perform computational tasks. Due to the relative simplicity of this approach the implementation in hardware is practicable, particularly the delay-based reservoir computing paradigm. Delay-based reservoirs use a single non-linear node subject to self-feedback. The high-dimensional dynamics that arise due to the feedback are utilised by driving the reservoir with time-multiplexed inputs. There have been a number of successful implementations of delay-based reservoirs in electronic, opto-electronic and photonic systems, among others. However, a challenge that remains is the efficient optimisation of a reservoir for performance on a variety of tasks. To this end we explore the influence of the delay-time on the performance of time-series prediction tasks and compare the computational performance of different methods of including task specific delay-timescales in a photonic reservoir setup.