Tag
This paper presents a general approach for quantifying local robustness in naive Bayes classifiers and generative forests, demonstrating its use as an indicator of prediction trustworthiness through perturbation analysis.
This paper proposes a method to extend factorized probability distributions defined on non-identical variable sets to a common measurable space, enabling principled comparison using distributional discrepancy measures.
This paper introduces lifted causal inference, leveraging parametric causal factor graphs to efficiently compute causal effects in relational domains, and presents the Lifted Causal Inference (LCI) algorithm for polynomial-time inference.
This paper presents diffusion models as part of a family of techniques that withhold information and train models to guess it, arguing that diffusion's destroying approach is flexible and advantageous, especially in data-scarce settings; it also discusses exploration problems and introduces a novel kind of probabilistic graphical model.
This paper revisits the theoretical foundations for detecting commutative factors in factor graphs, correcting a previously mistaken sufficient condition and presenting corrected algorithms.