particular the members of the Program Committee and all the additional external
reviewers for their time and effort in carefully reviewing and judging the submissions.
We further thank all authors for submitting their work to ARCS and presenting
accepted papers. The workshops were organized and coordinated by Carsten Trinitis,
the proceedings were compiled by Thilo Pionteck, and Gerald Krell, the website was
maintained by Markus Hoffmann. Thanks to all these individuals and all the many
other people who helped in the organization of ARCS 2018.
April 2018
Mladen Berekovic
Dirk Koch
Rainer Buchty
Heiko Hamann
VI
Preface
Organization
General Chair
Mladen Berekovic
Technische Universit
ät Braunschweig, Germany
Program Co-chairs
Rainer Buchty
Technische Universit
ät Braunschweig, Germany
Heiko Hamann
Universit
ät zu Lübeck, Germany
Dirk Koch
University of Manchester, UK
Workshop and Tutorial Chair
Carsten Trinitis
Technical University of Munich, Germany
Publicity Chair
Rainer Buchty
Technische Universit
ät Braunschweig, Germany
Publication Chair
Thilo Pionteck
Otto von Guericke University Magdeburg, Germany
Local Organization
Anna Jankowski
Technische Universit
ät Braunschweig, Germany
Web Chair
Markus Hoffmann
Karlsruhe Institute of Technology, Germany
Program Committee
Hamid Amiri
University of Tunis El Manar, Tunisia
Michael Beigl
Karlsruhe Institute of Technology, Germany
Mladen Berekovic
Technische Universit
ät Braunschweig, Germany
J
ürgen Brehm
Leibniz Universit
ät Hannover, Germany
Uwe Brinkschulte
Goethe-Universit
ät Frankfurt am Main, Germany
Rainer Buchty
Technische Universit
ät Braunschweig, Germany
Jo
ão M. P. Cardoso
Universidade do Porto, Portugal
Laura Carrington
San Diego Supercomputer Center/University of California,
USA
Martin Dan
ĕk
daiteq s.r.o., Czech Republic
Nikitas Dimopoulos
University of Victoria, Canada
Ahmed El-Mahdy
Egypt-Japan University of Science and Technology, Egypt
Dietmar Fey
Friedrich-Alexander-Universit
ät Erlangen-Nürnberg,
Germany
William Fornaciari
Politecnico di Milano, Italy
Roberto Giorgi
University of Siena, Italy
Daniel Gracia P
érez
Thales Research and Technology, France
Jan Haase
Universit
ät zu Lübeck, Germany
J
örg Hähner
University of Augsburg, Germany
Heiko Hamann
Universit
ät zu Lübeck, Germany
Andreas Herkersdorf
Technical University of Munich, Germany
Christian Hochberger
Technische Universit
ät Darmstadt, Germany
Koji Inoue
Kyushu University, Japan
Gert Jervan
Tallinn University of Technology, Estonia
Wolfgang Karl
Karlsruhe Institute of Technology, Germany
J
örg Keller
Fern Universit
ät in Hagen, Germany
Andreas Koch
Technische Universit
ät Darmstadt, Germany
Dirk Koch
University of Manchester, UK
Hana Kub
átová
FIT CTU, Prague, Czech Republic
Olaf Landsiedel
Chalmers University of Technology, Sweden
Erik Maehle
Universit
ät zu Lübeck, Germany
Alex Orailoglu
University of California, San Diego, USA
Luis Miguel Pinho
CISTER, ISEP, Portugal
Thilo Pionteck
Otto von Guericke University Magdeburg, Germany
Pascal Sainrat
IRIT, Universit
é de Toulouse, France
Luca Santinelli
ONERA, France
Toshinori Sato
Fukuoka University, Japan
Wolfgang
Schr
öder-Preikschat
Friedrich-Alexander-Universit
ät Erlangen-Nürnberg,
Germany
Martin Schulz
Technical University of Munich, Germany
Muhammad Sha
fique
Vienna University of Technology, Austria
Cristina Silvano
Politecnico di Milano, Italy
Leonel Sousa
Universidade de Lisboa, Portugal
Rainer G. Spallek
Technische Universit
ät Dresden, Germany
Olaf Spinczyk
Technische Universit
ät Dortmund, Germany
Benno Stabernack
Fraunhofer Institute for Telecommunications,
Heinrich Hertz Institute, Germany
Walter Stechele
Technical University of Munich, Germany
Djamshid Tavangarian
Universit
ät Rostock, Germany
J
ürgen Teich
Friedrich-Alexander-Universit
ät Erlangen-Nürnberg,
Germany
Sven Tomforde
University of Kassel, Germany
Eduardo Tovar
Polytechnic Institute of Porto, Portugal
Carsten Trinitis
Technical University of Munich, Germany
Nicolas Tsiftes
SICS Swedish ICT, Sweden
VIII
Organization
Sascha Uhrig
Airbus, Germany
Theo Ungerer
University of Augsburg, Germany
Hans Vandierendonck
Queen
’s University Belfast, UK
Stephane Vialle
CentraleSupelec and UMI GT-CNRS 2958, France
Lucian Vintan
Lucian Blaga University of Sibiu, Romania
Klaus Waldschmidt
Goethe-Universit
ät Frankfurt am Main, Germany
Dominik Wist
BIOTRONIC Berlin, Germany
Stephan Wong
Delft University of Technology, The Netherlands
Sungjoo Yoo
Seoul National University, South Korea
Additional Reviewers
Afzal, Ayesha
Becker, Thomas
Borghorst, Hendrik
Brand, Marcel
Bromberger, Michael
Buschhoff, Markus
Courtaud, C
édric
Dellagostin Souza, Jeckson
Eitschberger, Patrick
Freitag, Johannes
Frickenstein, Alexander
Friesel, Daniel
Gante, Jo
ão
Ghasempouri, Tara
Gottschling, Philip
Hoffmann, Markus
Hofmann, Jaco
Hoozemans, Joost
Joseph, Moritz
Jung, Lukas
Khalid, Faiq
Korinth, Jens
Lorenzon, Arthur
Martins, Paulo
Mische, J
örg
Niazmand, Behrad
Ozen, Elbruz
Payandeh Azad, Siavoosh
Perner, Cora
Procaccini, Marco
Pusz, Oskar
Reif, Stefan
Rheindt, Sven
Schirmeier, Horst
Schmaus, Florian
Schwarz, Alexander
Shuka, Romeo
Singh, Jasdeep
Sommer, Lukas
Spiekermann, Daniel
Srivatsa, Akshay
Stegmeier, Alexander
Terraneo, Federico
Tsiokanis, Ioannis
Wirsch, Ramon
Witterauf, Michael
W
ägemann, Peter
Yang, Dai
Zoni, Davide
Organization
IX
Biologically-Inspired
Massively-Parallel Computation
(Keynote Talk)
Steve Furber
Abstract.
The SpiNNaker project has delivered a massively-parallel computer
incorporating half a million ARM processor cores (with the ultimate goal of
expanding to a million cores) aimed at supporting large-scale models of spiking
neural systems that run in biological real time. The project has been 20 years in
conception and 10 years in construction, and the machine has been on line for a
year and a half as one of two major neuromorphic platforms supported by the
EU Flagship Human Brain Project.
The key problem to address in supporting biological neural networks is the
very high connectivity of those networks: each neuron typically receives con-
nections from, and connects to, around 10,000 (and sometimes as many as
250,000) other neurons. Biological neurons communicate primarily by issuing
action potentials or
“spikes”, and in SpiNNaker each spike is sent as a small
packet through a packet-switched fabric. The high connectivity is achieved by
using a multicast communication protocol. Each SpiNNaker chip incorporates
18 ARM968 processor cores and a packet router, where information about
packet destinations is held in tables in the router. This exploits the fact that the
topology of the biological network is static or, in the case of synaptogenesis or
neurogenesis, at most slowly changing.
The equations describing the neurons and the synapses (including learning
rules) are implemented on software running on the ARM processor cores, so the
problem of mapping a particular network application onto the machine has two
Do'stlaringiz bilan baham: |