Marginal Distribution
Marginal distributions, representing the probability distribution of a subset of variables within a larger system, are a central focus in various fields, with research aiming to efficiently estimate, reconstruct, and utilize them for diverse applications. Current research emphasizes developing novel algorithms and models, such as normalizing flows and iterative proportional fitting, to handle high-dimensional data and address challenges like privacy preservation and computational complexity in tasks ranging from cosmological parameter inference to synthetic data generation. The accurate and efficient handling of marginal distributions is crucial for advancing machine learning, statistical inference, and privacy-preserving data analysis, impacting fields from astronomy to social network analysis. Furthermore, understanding and leveraging marginal distributions is key to improving the accuracy and efficiency of various machine learning models.
Papers
Bounding the Excess Risk for Linear Models Trained on Marginal-Preserving, Differentially-Private, Synthetic Data
Yvonne Zhou, Mingyu Liang, Ivan Brugere, Dana Dachman-Soled, Danial Dervovic, Antigoni Polychroniadou, Min Wu
Variational Shapley Network: A Probabilistic Approach to Self-Explaining Shapley values with Uncertainty Quantification
Mert Ketenci, Iñigo Urteaga, Victor Alfonso Rodriguez, Noémie Elhadad, Adler Perotte