Implementing a CNN Universal Machine on FPGA: state-of-the-art and key challenges

Conference: ISTET 2009 - VXV International Symposium on Theoretical Engineering
06/22/2009 - 06/24/2009 at Lübeck, Germany

Proceedings: ISTET 2009

Pages: 5Language: englishTyp: PDF

Personal VDE Members are entitled to a 10% discount on this title

Schwarzlmüller, Christopher; Kyamakya, Kyandoghere (Alpen Adria University, Klagenfurt, Austria)

Cellular Neural Networks (CNN) is a massive computing paradigm which became very popular in the last decades. A Cellular Neural Network Universal Machine is an extension of the CNN concept. An implementation of CNN-UM on Field Programmable Gate Arrays (FPGA) appears attractive because their full computational power comes to a life only in hardware. Besides FPGA there are many different possibilities to implement a CNN-UM. The following questions will be answered while reading this paper: What is the CNN paradigm? Which application areas are of interest and what requirements are to meet? What is a CNN-UM? Which ways are possible to implement a CNN-UM – what are the differences? Which problems occur while implementing a CNN-UM on FPGA?