Scalable Conditional Independence Testing Using Spectral Representations

This research explores improvements in the efficiency and scalability of conditional independence testing, a crucial aspect of causal inference and machine learning. The use of spectral representations offers a novel approach to address computational bottlenecks in this important field.