Yaniv Tenzer

TL;DR: This work theoretically substantiates the conjecture that for many copula families the magnitude of Spearman's rank correlation coefficient is monotonic in the expected contribution of an edge in network, namely the negative copula entropy and suggests a novel Bayesian approach that makes use of a prior over values of spearman's rho for learning copula-based models that involve a mix of copula Families.

TL;DR: This work develops a test motivated by the classic Hoeffding's statistic, and uses two approaches to compute its distribution under the null: a bootstrap-based approach and an exact permutation-test with non-uniform probability of permutations.

TL;DR: This work defines the concept of a hypothetically optimal predictor of variable, and shows how it can be used to discover useful hidden variables in the expressive framework of copula networks, and demonstrates the merit of the approach for learning succinct models that generalize well in several real-life domains.

TL;DR: This work establishes a monotonic relationship between the mutual information and the copula dependence parameter, for a wide range of copula families, and gives rise to highly efficient proxy to the expected likelihood, which allows for scalable model selection.