Dependence Measure

Dependence measures quantify the statistical relationship between variables, serving as crucial tools in various fields including machine learning and time series analysis. Current research focuses on developing and benchmarking these measures for tasks like feature selection, mitigating confounding factors in complex datasets (e.g., medical imaging), and improving the efficiency of black-box explanation methods. These advancements are impacting diverse applications, from enhancing the reliability of deep learning models to enabling more accurate clustering of circular time series data, such as wind direction patterns. The development of universally consistent dependence measures further broadens their applicability across a range of statistical testing problems.