Research and Implementation of ε-SVR Training Method Based on FPGA

摘要: Online training of Support Vector Regression (SVR) in the field machine learning is a computationally complex algorithm. Due to need for multiple iterative processing training, SVR usually implemented on computer, and existing methods cannot be directly Field-Programmable Gate Array (FPGA), which restricts application range. This paper reconstructs framework implementation without precision loss reduce total latency required matrix update, reducing time consumption by 90%. A general e-SVR system with low Zynq platform. Taking regression samples two-dimensional as an example, maximum acceleration ratio 27.014× compared microcontroller platform energy 12.449% microcontroller. From experiments University California, Riverside (UCR) series data set. The results obtain excellent effects. minimum coefficient determination 0.996, running less than 30 ms, can meet requirements different applications real-time regression.