Fuzzy Logic-Based Adaptive Techniques for Learning Rate in Linear Regression Model

Abstract

Linear regression models performance based on the selection of the appropriate learning rate which is commonly used in machine learning applications. Conventional gradient descent methods often rely on constant learning rates, which may cause slow convergence or issues with divergence. To address this limitation, this study presents a fuzzy logic-driven adaptive learning rate selection method that employs Python coding. The proposed method adjusts the learning rate in real-time based on the error size, enabling quicker convergence and improving the model's precision. The experimental results show that integrating fuzzy logic into the learning rate selection process can accelerate convergence speed and enhance prediction accuracy, highlighting its ability to optimize the training of machine learning models.