Learning Bayesian networks from data: An information-theory based approach

This paper provides algorithms that use an information-theoretic analysis to learn Bayesian network structures from data. Based on our three-phase learning framework, we develop efficient algorithms that can effectively learn Bayesian networks, requiring only polynomial numbers of conditional independence (CI) tests in typical cases. We provide precise conditions that specify when these algorithms are guaranteed to be correct as well as empirical evidence (from real world applications and simulation tests) that demonstrates that these systems work efficiently and reliably in practice.

Kuznetsov independence for interval-valued expectations and sets of probability distributions: Properties and algorithms

Kuznetsov independence of variables X and Y means that, for any pair of bounded functions f(X) and g(Y), E[f(X)g(Y)]=E[f(X)] *times* E[g(Y)], where E[.] denotes interval-valued expectation and *times* denotes interval multiplication. We present properties of Kuznetsov independence for several variables, and connect it with other concepts of independence in the literature; in particular we show that strong extensions are always included in sets of probability distributions whose lower and upper expectations satisfy Kuznetsov independence. We introduce an algorithm that computes lower expectations subject to judgments of Kuznetsov independence by mixing column generation techniques with nonlinear programming. Finally, we define a concept of conditional Kuznetsov independence, and study its graphoid properties.

Conditional production of superpositions of coherent states with inefficient photon detection

It is shown that a linear superposition of two macroscopically distinguishable optical coherent states can be generated using a single photon source and simple all-optical operations. Weak squeezing on a single photon, beam mixing with an auxiliary coherent state, and photon detecting with imperfect threshold detectors are enough to generate a coherent state superposition in a free propagating optical field with a large coherent amplitude (alpha>2) and high fidelity (F>0.99). In contrast to all previous schemes to generate such a state, our scheme does not need photon number resolving measurements nor Kerr-type nonlinear interactions. Furthermore, it is robust to detection inefficiency and exhibits some resilience to photon production inefficiency.

Thickness independence of true phase transition temperatures in barium strontium titanate films

The functional properties of two types of barium strontium titanate (BST) thin film capacitor structures were studied: one set of structures was made using pulsed-laser deposition (PLD) and the other using chemical solution deposition. While initial observations on PLD films looking at the behavior of T-m (the temperature at which the maximum dielectric constant was observed) and T-c(*) (from Curie-Weiss analysis) suggested that the paraelectric-ferroelectric phase transition was progressively depressed in temperature as BST film thickness was reduced, further work suggested that this was not the case. Rather, it appears that the temperatures at which phase transitions occur in the thin films are independent of film thickness. Further, the fact that in many cases three transitions are observable, suggests that the sequence of symmetry transitions that occur in the thin films are the same as in bulk single crystals. This new observation could have implications for the validity of the theoretically produced thin film phase diagrams derived by Pertsev [Phys. Rev. Lett. 80, 1988 (1998)] and extended by Ban and Alpay [J. Appl. Phys. 91, 9288 (2002)]. In addition, the fact that T-m measured for virgin films does not correlate well with the inherent phase transition behavior, suggests that the use of T-m alone to infer information about the thermodynamics of thin film capacitor behavior, may not be sufficient. (C) 2004 American Institute of Physics.