Graphon Cross-Validation: Assessing Models on Network Data

Explore a comprehensive conference talk on graphon cross-validation techniques for assessing models on network data. Delve into the challenges of overfitting in graphon models and the importance of accurate parameter calibration. Learn about a novel graphon validation method designed to overcome limitations of existing approaches, improving both practical implementation and theoretical understanding. Examine the results of extensive simulations and real-world experiments demonstrating the superior computational efficiency and accuracy of the proposed method. Gain insights into the asymptotic convergence of the optimal model identified through this approach to the ideal model determined by Kullback-Leibler divergence. Discover how this theoretical advancement enhances the applicability of network analysis across diverse settings.

Wenxuan Zhong: Graphon Cross-Validation: Assessing Models on Network Data #ICBS2024