Paper Accepted into IOPscience Neuromorphic Computing and Engineering

We are proud to announce that the article 'Distributed representations enable robust multi-timescale symbolic computation in neuromorphic hardware' was accepted into the IOPscience journal Neuromorphic Computing and Engineering. This journal is a multidisciplinary, open access journal publishing cutting edge research on the design, development and application of artificial neural networks and systems from both a neuromorphic hardware and computational perspective.

The paper, written by Madison Cotteret and co-authored by Elisabetta Chicca and Hugh Greatorex together with colleagues from the UZH and ETH Zurich, Switzerland, from the Tsinghua University, China, from the Forschungszentrum Jülich, Germany and from the Kiel University, Germany presents a novel method for programming recurrent spiking neural networks (RSNNs) to perform robust multi-timescale computation. By leveraging high-dimensional distributed representations, the research introduces a single-shot weight learning scheme that directly embeds finite state machines into RSNN dynamics. Results showed that the method scales effectively to large state machines without requiring extensive parameter fine-tuning or hardware-specific optimization. This work highlights the potential of distributed symbolic representations as a powerful framework for embedding cognitive algorithms into neuromorphic hardware.