Factor Graph Inference Engine on the SpiNNaker Neural Computing System

  • Indar Sugiarto
  • Jörg Conradt
Part of the Lecture Notes in Computer Science book series (LNCS, volume 8681)


This paper presents a novel method for implementing Factor Graphs in a SpiNNaker neural computing system. The SpiNNaker system provides resources for fine-grained parallelism, designed for implementing a distributed computing system. We present a framework which utilizes available SpiNNaker resources to implement a discrete Factor Graph: a powerful graphical model for probabilistic inference. Our framework allows mapping and routing a Factor Graph on the SpiNNaker hardware using SpiNNaker’s event-based communication system. An example application of the proposed framework in a real-world robotics scenario is given and the result shows that the framework can handle computation of 26.14 MFLOPS only in 30.5ms. We demonstrate that the framework easily extends for larger Factor Graph networks in a bigger SpiNNaker system, which makes it suitable for complex and challenging computational intelligence tasks.


parallel distributed system SpiNNaker Factor Graph 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Kschischang, F.R., Frey, B.J., Loeliger, H.-A.: Factor Graphs and the Sum-Product Algorithm. IEEE Transactions on Information Theory 47(2), 498–519 (2001)CrossRefzbMATHMathSciNetGoogle Scholar
  2. 2.
    Frey, B.J.: Extending Factor Graphs so as to Unify Directed and Undirected Graph-ical Models. In: Proc. The 19th Conference on Uncertainty in Artificial Intelligence (2003)Google Scholar
  3. 3.
    Abbeel, P., Koller, D., Andrew, Y.N.: Learning Factor Graphs in Polynomial Time & Sample Complexity. In: Proc. the 21th Conference on Uncertainty in Artificial Intelligence, UAI 2005 (2005)Google Scholar
  4. 4.
    Khan, M., Lester, D., Plana, L.A., Rast, A., Jin, X., Painkras, E., Furber, S.B.: SpiNNaker: Mapping Neural Networks onto a Massively-Parallel Chip Multiprocessor. In: Proc. IEEE International Joint Conference on Neural Networks (IJCNN), pp. 2849–2856 (2008)Google Scholar
  5. 5.
    Plana, L.A., Bainbridge, J., Furber, S., Salisbury, S., Shi, Y., Wu, J.: An on-Chip and Inter-Chip Communications Network for the SpiNNaker Massively-Parallel Neural Net Simulator. In: Proc. 2nd ACM/IEEE NoCS, pp. 215–216. IEEE (2008)Google Scholar
  6. 6.
    Galluppi, F., Davies, S., Rast, A., Sharp, T., Plana, L.A., Furber, S.: A Hierarchical Configuration System for a Massively Parallel Neural Hardware Platform. In: Proc. 9th Conference of Computing Frontiers, pp. 183–192. ACM, New York (2012)Google Scholar
  7. 7.
    Sugiarto, I., Maier, P., Conradt, J.: Reasoning with discrete factor graph. In: Proc. International Conference on Robotics, Biomimetics, Intelligent Computational Systems 2013 (Robionetics), Yogyakarta, Indonesia (2013)Google Scholar
  8. 8.
    Sugiarto, I., Conradt, J.: Discrete belief propagation network using population coding and factor graph for kinematic control of a mobile robot. In: Proc. International Conference on Computational Intelligence and Cybernetics 2013 (Cyberneticscom 2013), Indonesia,Google Scholar
  9. 9.
    Denk, C., Llobet-Blandino, F., Gallupi, F., Plana, L., Furber, S., Conradt, J.: Real-Time In-terface Board for Closed-Loop Robotic Tasks on the SpiNNaker Neural Computing System. In: Proc. International Conference on Artificial Neural Networks (ICANN), Sofia, Bulgaria, pp. 467–474 (2013)Google Scholar
  10. 10.
    Lin, M., Lebedev, I., Wawrzynek, J.: High-Throughput Bayesian Computing Machine with Reconfigurable Hardware. In: Proc. The 18th Annual ACM/SIGDA International Symposium on Field Programmable Gate Arrays (FPGA 210), Monterey, USA, pp. 73–82 (2010)Google Scholar

Copyright information

© Springer International Publishing Switzerland 2014

Authors and Affiliations

  • Indar Sugiarto
    • 1
  • Jörg Conradt
    • 1
  1. 1.Neuroscientific System Theory, Fakultät für Elektro- und InformationstechnikTechnische Universität MünchenGermany

Personalised recommendations