Learning with Graphical Models

Probabilistic graphical models are being used widely in artiicial intelligence, for instance, in diagnosis and expert systems, as a uniied qualitative and quantitative framework for representing and reasoning with probabilities and independencies. Their development and use spans several elds including artiicial intelligence, decision theory and statistics, and provides an important bridge between these communities. This paper explains how these models can be extended to machine learning, neural networks, knowledge discovery, and knowledge reenement. This provides a uniied framework that combines lessons learned from the connec-tionist, artiicial intelligence, and statistical communities, and provides a smooth transition between the inference found in diagnostic and model-based systems and that found in learning systems. This also ooers a set of principles for developing a learning toolbox , or more ambitiously , a learning compiler, whereby a learning or discovery system can be compiled from speciications. Many of the popular learning algorithms can be compiled in this way from graphical speciications. While in a sense this paper is a multidisciplinary review of learning, the main contribution here is the presentation of the material within the unifying framework of graphical models, and the observation that, as a result, the process of developing learning algorithms can be partly automated.