Abstract-In this paper, we propose a new approach aiming to ameliorate the performances of the regularization networks (RN) method and speed up its computation time. A considerable rapidity in totaling calculation time and high performance were accomplished through conveying difficult calculation charges to FPGA. Using Xilinx System Generator, a successful HW/SW CoDesign was constructed to accelerate the Gramian matrix computation. Experimental results involving two real data sets of Wiener-Hammerstein benchmark with process noise prove the efficiency of the approach. The implementation results demonstrate the efficiency of the heterogeneous architecture, presenting a speed-up factor of 40-50 orders of time, comparing to the CPU simulation.

Abstract-This work resumes the previous implementations of Support Vector Machine for Classification and Regression and explicates the different methods and approaches adopted. Ever since the rarity of works in the field of nonlinear systems regression, an implementation of testing phase of SVM was proposed exploiting the parallelism and reconfigurability of FieldProgrammable Gate Arrays (FPGA) platform. The nonlinear system chosen for application was a real challenging model: a fluid level control system existing in our laboratory. The implemented design with fixed point precision demonstrates good enough results comparing with the software performances based on the Normalized Mean Squared Error. Whereas, in term of computation time, a speed-up factor of 60 orders of time comparing to MATLAB results was achieved. Due to the flexibility of Xilinx System Generator, the design is capable to be reused for any other system with different data sets sizes and various kernel functions.

No abstract