New Advances and Theoretical Insights into EDML
نویسندگان
چکیده
EDML is a recently proposed algorithm for learning MAP parameters in Bayesian networks. In this paper, we present a number of new advances and insights on the EDML algorithm. First, we provide the multivalued extension of EDML, originally proposed for Bayesian networks over binary variables. Next, we identify a simplified characterization of EDML that further implies a simple fixed-point algorithm for the convex optimization problem that underlies it. This characterization further reveals a connection between EDML and EM: a fixed point of EDML is a fixed point of EM, and vice versa. We thus identify also a new characterization of EM fixed points, but in the semantics of EDML. Finally, we propose a hybrid EDML/EM algorithm that takes advantage of the improved empirical convergence behavior of EDML, while maintaining the monotonic improvement property of EM.
منابع مشابه
New insights into the prevention and treatment strategies for hypertrophic scars and keloids
Hypertrophic scars and keloids are fibrosis abnormalities associated with the accumulation of collagen and extra cellular matrix components. These scars are caused by abnormal wound healing, which may occur after skin injuries caused by surgery, trauma, burns, etc. and may have a large impact on the patients’ quality of life. Hypertrophic scars and colloids in addition to aesthetic problems can...
متن کاملEDML for Learning Parameters in Directed and Undirected Graphical Models
EDML is a recently proposed algorithm for learning parameters in Bayesian networks. It was originally derived in terms of approximate inference on a meta-network which underlies the Bayesian approach to parameter estimation. While this initial derivation helped discover EDML in the first place and provided a concrete context for identifying some of its properties (e.g., in contrast to EM), the ...
متن کاملInstruments and Methods Distribution of air bubbles in the EDML and EDC (Antarctica) ice cores, using a new method of automatic image analysis
Air bubbles in ice cores play an essential role in climate research, not only because they contain samples of the palaeoatmosphere, but also because their shape, size and distribution provide information about the past firn structure and the embedding of climate records into deep ice cores. In this context, we present profiles of average bubble size and bubble number for the entire EDML (Antarc...
متن کاملTheoretical insights into the adsorption behavior of CO molecules on the pure and Vn-doped BN nanotubes
Interaction of pure and Vn-doped (8, 0), (12, 0) and (16, 0) boron nitride nanotubes with CO molecules was studied using B3LYP/6-311++G(d) theoretical level. Substituting V instead of B atoms, increased the reactivity of nanotube. From the results, the complex stability depends on the direction and the number of the CO molecules interacted with the nanotube. In this work, the quantum...
متن کاملEFL Pronunciation Teaching: A Theoretical Review
This study aims to represent the developing status of pronunciation teaching and presents the current perspectives on pronunciation learning and teaching, coupled with innovative approaches and techniques/activities. It is argued that pronunciation teaching methodologies have changed over decades since the Reform Movement. The exact status of teaching pronunciation appeared first in the Audio L...
متن کامل