A Mott Insulator-Based Oscillator Circuit for Reservoir Computing

IEEE International Symposium on Circuits and Systems proceedings, pp. 1–5

Abstract

In this work, we use a Cr-doped V2O3based Mott oscillator circuit to build a reservoir computing system that has much smaller model size than an equivalent LSTM. In contrast to an LSTM, our reservoir computing system can be trained very efficiently and on-line, with very low latency. We demonstrate close to state-of-the-art performance with three benchmark tasks: speech recognition, handwritten digit recognition, and HDD channel decoding. We show that our Mott circuit-based reservoir computing system brings significant reduction in power consumption and inference speed compared to CPU, GPU, or FPGA based systems.

Authors 11

1. Wen Ma

   Western Digital (United States)

   Affiliation as printed

   Western Digital Research, San Jose, CA 95119, USA

2. Tyler Hennen Aachen IWE II

   RWTH Aachen University

   Affiliation as printed

   IWE II, RWTH Aachen University, 52074 Aachen, Germany

3. Martin Lueker-Boden

   Western Digital (United States)

   Affiliation as printed

   Western Digital Research, San Jose, CA 95119, USA

4. Rick Galbraith

   Western Digital (United States)

   Affiliation as printed

   Western Digital HDD R&D, Rochester, MN 55901

5. Jonas Goode

   Western Digital (United States)

   Affiliation as printed

   Western Digital HDD R&D, Rochester, MN 55901

6. Won Ho Choi

   Western Digital (United States)

   Affiliation as printed

   Western Digital Research, San Jose, CA 95119, USA

7. Pi-Feng Chiu

   Western Digital (United States)

   Affiliation as printed

   Western Digital Research, San Jose, CA 95119, USA

8. Jonathan A. J. Rupp Aachen IWE II

   RWTH Aachen University

   Affiliation as printed

   IWE II, RWTH Aachen University, 52074 Aachen, Germany

9. Dirk J. Wouters Aachen IWE II

   RWTH Aachen University

   Affiliation as printed

   IWE II, RWTH Aachen University, 52074 Aachen, Germany

10. Rainer M. Waser Aachen IWE II

    RWTH Aachen University

    Affiliation as printed

    IWE II, RWTH Aachen University, 52074 Aachen, Germany; Peter Grünberg Institute, Forschungszentrum Jülich GmbH, 52428 Jülich, Germany

11. Daniel Bedau

    Western Digital (United States)

    Affiliation as printed

    Western Digital Research, San Jose, CA 95119, USA