Meta Tree

Meta-trees represent a novel approach to improve decision-making in machine learning, addressing limitations of traditional decision trees by offering enhanced predictive performance and preventing overfitting. Current research focuses on developing ensemble methods, such as boosting algorithms, to construct multiple meta-trees and on integrating meta-trees with graph neural networks and variational autoencoders for handling complex relational data, particularly in heterogeneous information networks. This work aims to improve the efficiency and accuracy of machine learning models across various applications, from webpage classification to university ranking, by leveraging the inherent structure and relationships within data.