Machine Learning Regression Model on FPGA
Majid Pakdel · 2025
This chapter presents the implementation of a machine learning regression model on FPGA using MATLAB, Simulink, and Vivado. The study utilizes an admission dataset from Kaggle to predict university admission chances based on multiple input parameters. Initially, a linear regression model is trained in MATLAB using the least squares method, and its performance is evaluated through mean square error and prediction plots. The trained model is then converted to fixed-point representation and implemented in Simulink, facilitating the generation of synthesizable VHDL code using HDL Coder. The generated VHDL code is imported into Vivado for synthesis and behavioral simulation. The results demonstrate that the FPGA-based implementation closely replicates MATLAB predictions, confirming the feasibility of deploying machine learning regression models on hardware platforms for real-time applications.