Modal Regression

Modal regression focuses on estimating the most frequent value (mode) of a dependent variable given independent variables, offering robustness to outliers and noise compared to traditional mean-based regression. Current research emphasizes developing efficient and scalable methods for multi-modal data, employing architectures like Mixture of Experts (MoE) and incorporating pre-trained uni-modal models to improve performance and generalization. This work is significant for its applications in diverse fields such as computer vision (object tracking), wind speed forecasting, and transportation systems modeling, where handling multiple data sources and noisy data is crucial. Furthermore, theoretical advancements are improving our understanding of modal regression's behavior under various data dependencies.