Pattern representation and recognition with accelerated analog neuromorphic systems

Mihai A. Petrovici, Sebastian Schmitt, Johann Klähn, David Stöckel, Anna Schroeder, Guillaume Bellec, Johannes Bill, Oliver Breitwieser, Ilja Bytschok, Andreas Grübl, Maurice Güttler, Andreas Hartel, Stephan Hartmann, Dan Husmann, Kai Husmann, Sebastian Jeltsch, Vitali Karasenko, Mitja Kleider, Christoph Koke, Alexander Kononov & 15 others Christian Mauch, Paul Müller, Johannes Partzsch, Thomas Pfeil, Stefan Schiefer, Stefan Scholze, Anand Subramoney, Vasilis Thanasoulis, Bernhard Vogginger, Robert Legenstein, Wolfgang Maass, René Schüffny, Christian Mayr, Johannes Schemmel, Karlheinz Meier

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Despite being originally inspired by the central nervous system, artificial neural networks have diverged from their biological archetypes as they have been remodeled to fit particular tasks. In this paper, we review several possibilites to reverse map these architectures to biologically more realistic spiking networks with the aim of emulating them on fast, low-power neuromorphic hardware. Since many of these devices employ analog components, which cannot be perfectly controlled, finding ways to compensate for the resulting effects represents a key challenge. Here, we discuss three different strategies to address this problem: the addition of auxiliary network components for stabilizing activity, the utilization of inherently robust architectures and a training method for hardware-emulated networks that functions without perfect knowledge of the system's dynamics and parameters. For all three scenarios, we corroborate our theoretical considerations with experimental results on accelerated analog neuromorphic platforms.
Original languageEnglish
Number of pages4
JournalarXiv.org e-Print archive
Volumepreprint arXiv:1703.06043
DOIs
Publication statusPublished - 17 Mar 2017

Fingerprint

Hardware
Network components
Neurology
Dynamical systems
Neural networks

Keywords

  • q-bio.NC
  • cs.NE
  • stat.ML

Fields of Expertise

  • Information, Communication & Computing

Cite this

Petrovici, M. A., Schmitt, S., Klähn, J., Stöckel, D., Schroeder, A., Bellec, G., ... Meier, K. (2017). Pattern representation and recognition with accelerated analog neuromorphic systems. arXiv.org e-Print archive, preprint arXiv:1703.06043. https://doi.org/10.1109/ISCAS.2017.8050530

Pattern representation and recognition with accelerated analog neuromorphic systems. / Petrovici, Mihai A.; Schmitt, Sebastian; Klähn, Johann; Stöckel, David; Schroeder, Anna; Bellec, Guillaume; Bill, Johannes; Breitwieser, Oliver; Bytschok, Ilja; Grübl, Andreas; Güttler, Maurice; Hartel, Andreas; Hartmann, Stephan; Husmann, Dan; Husmann, Kai; Jeltsch, Sebastian; Karasenko, Vitali; Kleider, Mitja; Koke, Christoph; Kononov, Alexander; Mauch, Christian; Müller, Paul; Partzsch, Johannes; Pfeil, Thomas; Schiefer, Stefan; Scholze, Stefan; Subramoney, Anand; Thanasoulis, Vasilis; Vogginger, Bernhard; Legenstein, Robert; Maass, Wolfgang; Schüffny, René; Mayr, Christian; Schemmel, Johannes; Meier, Karlheinz.

In: arXiv.org e-Print archive, Vol. preprint arXiv:1703.06043, 17.03.2017.

Research output: Contribution to journalArticleResearchpeer-review

Petrovici, MA, Schmitt, S, Klähn, J, Stöckel, D, Schroeder, A, Bellec, G, Bill, J, Breitwieser, O, Bytschok, I, Grübl, A, Güttler, M, Hartel, A, Hartmann, S, Husmann, D, Husmann, K, Jeltsch, S, Karasenko, V, Kleider, M, Koke, C, Kononov, A, Mauch, C, Müller, P, Partzsch, J, Pfeil, T, Schiefer, S, Scholze, S, Subramoney, A, Thanasoulis, V, Vogginger, B, Legenstein, R, Maass, W, Schüffny, R, Mayr, C, Schemmel, J & Meier, K 2017, 'Pattern representation and recognition with accelerated analog neuromorphic systems' arXiv.org e-Print archive, vol. preprint arXiv:1703.06043. https://doi.org/10.1109/ISCAS.2017.8050530
Petrovici MA, Schmitt S, Klähn J, Stöckel D, Schroeder A, Bellec G et al. Pattern representation and recognition with accelerated analog neuromorphic systems. arXiv.org e-Print archive. 2017 Mar 17;preprint arXiv:1703.06043. https://doi.org/10.1109/ISCAS.2017.8050530
Petrovici, Mihai A. ; Schmitt, Sebastian ; Klähn, Johann ; Stöckel, David ; Schroeder, Anna ; Bellec, Guillaume ; Bill, Johannes ; Breitwieser, Oliver ; Bytschok, Ilja ; Grübl, Andreas ; Güttler, Maurice ; Hartel, Andreas ; Hartmann, Stephan ; Husmann, Dan ; Husmann, Kai ; Jeltsch, Sebastian ; Karasenko, Vitali ; Kleider, Mitja ; Koke, Christoph ; Kononov, Alexander ; Mauch, Christian ; Müller, Paul ; Partzsch, Johannes ; Pfeil, Thomas ; Schiefer, Stefan ; Scholze, Stefan ; Subramoney, Anand ; Thanasoulis, Vasilis ; Vogginger, Bernhard ; Legenstein, Robert ; Maass, Wolfgang ; Schüffny, René ; Mayr, Christian ; Schemmel, Johannes ; Meier, Karlheinz. / Pattern representation and recognition with accelerated analog neuromorphic systems. In: arXiv.org e-Print archive. 2017 ; Vol. preprint arXiv:1703.06043.
@article{1984b254a9474572abe4261bf5f7699b,
title = "Pattern representation and recognition with accelerated analog neuromorphic systems",
abstract = "Despite being originally inspired by the central nervous system, artificial neural networks have diverged from their biological archetypes as they have been remodeled to fit particular tasks. In this paper, we review several possibilites to reverse map these architectures to biologically more realistic spiking networks with the aim of emulating them on fast, low-power neuromorphic hardware. Since many of these devices employ analog components, which cannot be perfectly controlled, finding ways to compensate for the resulting effects represents a key challenge. Here, we discuss three different strategies to address this problem: the addition of auxiliary network components for stabilizing activity, the utilization of inherently robust architectures and a training method for hardware-emulated networks that functions without perfect knowledge of the system's dynamics and parameters. For all three scenarios, we corroborate our theoretical considerations with experimental results on accelerated analog neuromorphic platforms.",
keywords = "q-bio.NC, cs.NE, stat.ML",
author = "Petrovici, {Mihai A.} and Sebastian Schmitt and Johann Kl{\"a}hn and David St{\"o}ckel and Anna Schroeder and Guillaume Bellec and Johannes Bill and Oliver Breitwieser and Ilja Bytschok and Andreas Gr{\"u}bl and Maurice G{\"u}ttler and Andreas Hartel and Stephan Hartmann and Dan Husmann and Kai Husmann and Sebastian Jeltsch and Vitali Karasenko and Mitja Kleider and Christoph Koke and Alexander Kononov and Christian Mauch and Paul M{\"u}ller and Johannes Partzsch and Thomas Pfeil and Stefan Schiefer and Stefan Scholze and Anand Subramoney and Vasilis Thanasoulis and Bernhard Vogginger and Robert Legenstein and Wolfgang Maass and Ren{\'e} Sch{\"u}ffny and Christian Mayr and Johannes Schemmel and Karlheinz Meier",
note = "accepted at ISCAS 2017",
year = "2017",
month = "3",
day = "17",
doi = "10.1109/ISCAS.2017.8050530",
language = "English",
volume = "preprint arXiv:1703.06043",
journal = "arXiv.org e-Print archive",
publisher = "Cornell University Library",

}

TY - JOUR

T1 - Pattern representation and recognition with accelerated analog neuromorphic systems

AU - Petrovici, Mihai A.

AU - Schmitt, Sebastian

AU - Klähn, Johann

AU - Stöckel, David

AU - Schroeder, Anna

AU - Bellec, Guillaume

AU - Bill, Johannes

AU - Breitwieser, Oliver

AU - Bytschok, Ilja

AU - Grübl, Andreas

AU - Güttler, Maurice

AU - Hartel, Andreas

AU - Hartmann, Stephan

AU - Husmann, Dan

AU - Husmann, Kai

AU - Jeltsch, Sebastian

AU - Karasenko, Vitali

AU - Kleider, Mitja

AU - Koke, Christoph

AU - Kononov, Alexander

AU - Mauch, Christian

AU - Müller, Paul

AU - Partzsch, Johannes

AU - Pfeil, Thomas

AU - Schiefer, Stefan

AU - Scholze, Stefan

AU - Subramoney, Anand

AU - Thanasoulis, Vasilis

AU - Vogginger, Bernhard

AU - Legenstein, Robert

AU - Maass, Wolfgang

AU - Schüffny, René

AU - Mayr, Christian

AU - Schemmel, Johannes

AU - Meier, Karlheinz

N1 - accepted at ISCAS 2017

PY - 2017/3/17

Y1 - 2017/3/17

N2 - Despite being originally inspired by the central nervous system, artificial neural networks have diverged from their biological archetypes as they have been remodeled to fit particular tasks. In this paper, we review several possibilites to reverse map these architectures to biologically more realistic spiking networks with the aim of emulating them on fast, low-power neuromorphic hardware. Since many of these devices employ analog components, which cannot be perfectly controlled, finding ways to compensate for the resulting effects represents a key challenge. Here, we discuss three different strategies to address this problem: the addition of auxiliary network components for stabilizing activity, the utilization of inherently robust architectures and a training method for hardware-emulated networks that functions without perfect knowledge of the system's dynamics and parameters. For all three scenarios, we corroborate our theoretical considerations with experimental results on accelerated analog neuromorphic platforms.

AB - Despite being originally inspired by the central nervous system, artificial neural networks have diverged from their biological archetypes as they have been remodeled to fit particular tasks. In this paper, we review several possibilites to reverse map these architectures to biologically more realistic spiking networks with the aim of emulating them on fast, low-power neuromorphic hardware. Since many of these devices employ analog components, which cannot be perfectly controlled, finding ways to compensate for the resulting effects represents a key challenge. Here, we discuss three different strategies to address this problem: the addition of auxiliary network components for stabilizing activity, the utilization of inherently robust architectures and a training method for hardware-emulated networks that functions without perfect knowledge of the system's dynamics and parameters. For all three scenarios, we corroborate our theoretical considerations with experimental results on accelerated analog neuromorphic platforms.

KW - q-bio.NC

KW - cs.NE

KW - stat.ML

U2 - 10.1109/ISCAS.2017.8050530

DO - 10.1109/ISCAS.2017.8050530

M3 - Article

VL - preprint arXiv:1703.06043

JO - arXiv.org e-Print archive

JF - arXiv.org e-Print archive

ER -