What are Bayesian networks and why are their applications growing across all fields ? BY aDnan DaRWiChE

Bayesian networks problems that span across domains such as computer vision, the Web, and medical diagnosis. So what are Bayesian networks, and why are they widely used, either directly or indirectly, across so many fields and application areas? Intuitively, Bayesian networks provide a systematic and localized method for structuring probabilistic information about a situation into a coherent whole. They also provide a suite of algorithms that allow one to automatically derive many implications of this information, which can form the basis for important conclusions and decisions about the corresponding situation (for example, computing the overall reliability of a system, finding the most likely message that was sent across a noisy channel, identifying the most likely users that would respond to an ad, restoring a noisy image, mapping genes onto a chromosome, among others). Technically speaking, a Bayesian network is a compact representation of a probability distribution that is usually too large to be handled using traditional specifications from probability and statistics such as tables and equations. For example, Bayesian networks with thousands of variables have been constructed and reasoned about successfully, allowing one to efficiently represent and reason about probability distributions whose size is exponential in that number of variables (for example, in genetic linkBaYesian netWorKs HaVe been receiving considerable attention over the last few decades from scientists and engineers across a number of fields, including computer science, cognitive science, statistics, and philosophy. In computer science, the development of Bayesian networks was driven by research in artificial intelligence, which aimed at producing a practical framework for commonsense reasoning. Statisticians have also contributed to the development of Bayesian networks, where they are studied under the broader umbrella of probabilistic graphical models.5,11 Interestingly enough, a number of other more specialized fields, such as genetic linkage analysis, speech recognition, information theory and reliability analysis, have developed representations that can be thought of as concrete instantiations or restricted cases of Bayesian networks. For example, pedigrees and their associated phenotype/genotype information, reliability block diagrams, and hidden Markov models (used in many fields including speech recognition and bioinformatics) can all be viewed as Bayesian networks. Canonical instances of Bayesian networks also exist and have been used to solve standard key insights