Search:
Match:
2 results

Analysis

This paper addresses a limitation in Bayesian regression models, specifically the assumption of independent regression coefficients. By introducing the orthant normal distribution, the authors enable structured prior dependence in the Bayesian elastic net, offering greater modeling flexibility. The paper's contribution lies in providing a new link between penalized optimization and regression priors, and in developing a computationally efficient Gibbs sampling method to overcome the challenge of an intractable normalizing constant. The paper demonstrates the benefits of this approach through simulations and a real-world data example.
Reference

The paper introduces the orthant normal distribution in its general form and shows how it can be used to structure prior dependence in the Bayesian elastic net regression model.

Analysis

This paper addresses a fundamental question in the study of random walks confined to multidimensional spaces. The finiteness of a specific group of transformations is crucial for applying techniques to compute generating functions, which are essential for analyzing these walks. The paper provides new results on characterizing the conditions under which this group is finite, offering valuable insights for researchers working on these types of problems. The complete characterization in 2D and the constraints on higher dimensions are significant contributions.
Reference

The paper provides a complete characterization of the weight parameters that yield a finite group in two dimensions.