Hardware-Software Co-Design for Kohonen's Self-Organizing Map

Kohonen's self-organizing map (SOM) is a widely used technique to cluster unstructured data. It has applications in computer graphics, image processing, robotics, soft-computing and many more. The exact speci cation and time requirements may vary according to the concrete application, therefore a re-design of Kohonen's SOM o ering several di erent performance/cost trade-o s by using special purpose hardware acceleration would be bene cial. This paper introduces the main principles of systemlevel synthesis (SLS) which aims at constructing optimal hardware-software systems ful lling a high-level speci cation and applies its methodologies to construct di erent hardware-software SOM realizations corresponding to the concrete expectations.

[1]  F. Faggin,et al.  Neural network hardware , 1992, [Proceedings 1992] IJCNN International Joint Conference on Neural Networks.

[2]  Péter Arató,et al.  High Level Synthesis of Pipelined Datapaths , 2001 .

[3]  Z. Mann.  Optimization problems in system-level synthesis∗ , 2012 .

[4]  M. Chinnadurai,et al.  HIGH LEVEL SYNTHESIS , 2011 .

[5]  Teuvo Kohonen,et al.  Self-organized formation of topologically correct feature maps , 2004, Biological Cybernetics.

[6]  Minh N. Do,et al.  Youn-Long Steve Lin , 1992 .

[7]  Wolfgang Rosenstiel,et al.  A hardware supported system for a special online application of self-organizing map , 1996, Proceedings of International Conference on Neural Networks (ICNN'96).