A Survey of Task-driven Heterogeneous Feature Embedding and Selection
Published in Accepted in QE, Computer Science Department, Hong Kong Baptist University, 2023
As the complexity of real-world tasks and data sources continues to grow, the need for heterogeneous feature processing techniques becomes increasingly apparent. Traditional machine learning algorithms which is designed for homogeneous data can be degraded for heterogeneous data. How to handle heterogeneous features is a very important problem. Our research problem for handling heterogeneous data is divided into two sub-problem: heterogeneous feature embedding and heterogeneous feature selection. This survey explores the existing works for both the heterogeneous embedding algorithms and feature selection algorithms which is a very critical step in data analysis. The survey presents our methodologies and progress in addressing the identified problem, while outlining future plans.