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

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 37: Biopolymers, Biomaterials and Bioinspired Functional Materials (joint session CPP/BP)

CPP 37.7: Talk

Thursday, March 21, 2024, 11:30–11:45, H 0111

Reservoir computing with organic fiber networks — •Richard Kantelberg, Anton Weissbach, Peter Steiner, Peter Birkholz, Hans Kleemann, and Karl Leo — Technische Universität Dresden, Dresden, Germany

Reservoir computing (RC) is a promising paradigm for machine learning that utilizes dynamic systems, known as reservoirs, to process and analyze complex temporal data. Organic mixed ionic electronic conductors (OMIECs) have emerged as a novel class of materials with intriguing properties, such as their ability to exhibit both electronic and ionic conductivity, as well as their biocompatibility, flexibility, and low power consumption. These features make OMIECs particularly suitable for the development of unconventional computing architectures. Conducting fiber networks grown by field-directed polymerization have been proven to be a suitable candidate for, e.g., heartbeat or image classification tasks. However, the dependency between classification accuracy and device parameters is still rather unclear. We present recent findings interlinking electronic conductivity, system nonlinearity and reservoir size with the neuromorphic functionality and RC performance. The recent progress in reservoir computing using organic mixed ionic electronic provides valuable knowledge for the targeted development of fiber reservoirs. Given these findings, we are confident to further increase the classification accuracy by adopting the system to specific application scenarios, paving the way to future commercialization.

Keywords: Neuromorphic; Organic; Network; Machine Learning; OECT