Implementation Method of SVR Algorithm in Resource-Constrained Platform

With the development of the Internet of Things and edge computing, machine learning algorithms need to be deployed on resource-constrained embedded platforms. Support Vector Regression (SVR) is one of the most popular algorithms widely used in solving problems characterized by small samples, high-dimensional, and nonlinear, with its good generalization ability and prediction performance. However, SVR algorithm requires a lot of resources when it is implemented. Therefore, this paper proposes a method to implement SVR algorithm in the resource-constrained embedded platform. The method analyses the characteristics of the data in the SVR algorithm and the solution process of the algorithm. Then, according to the characteristics of the embedded platform, the implementation process of the algorithm is optimized. Experiments using UCI datasets show that the implemented SVR algorithm is correct and effective, and the optimized SVR algorithm reduces time and memory consumption at the same time, which is of great significance for the implementation of SVR algorithm in resource-constrained embedded platforms.