Neural Computation · 1997 · 260 citations · 43 references
ComputingComputer ScienceArtificial IntelligenceMachine LearningSensor FusionHidden Markov ModelsMarkov ChainsMarkov ProcessesPattern RecognitionSpeech RecognitionStatistical InferenceGraphical ModelsInference In Hidden Markov ModelsSpeech Pattern RecognitionTransition ProbabilitiesViterbi AlgorithmViterbi Decoding
Graphical techniques for modeling dependencies of random variables have been studied across fields such as statistics, physics, AI, speech recognition, image processing, and genetics, with formal frameworks developed largely independently. The paper investigates hidden Markov models and related structures within the framework of probabilistic independence networks. The authors review probabilistic independence networks and show that HMM inference algorithms are special cases of general PIN inference, providing inference and estimation tools for richer graphical models. It is shown that forward‑backward and Viterbi algorithms are special cases of general PIN inference, and the authors illustrate the advantages of this general approach with examples for sensor fusion and speech‑recognition coarticulation.
Graphical techniques for modeling the dependencies of random variables have been explored in a variety of different areas, including statistics, statistical physics, artificial intelligence, speech recognition, image processing, and genetics. Formalisms for manipulating these models have been developed relatively independently in these research communities. In this paper we explore hidden Markov models (HMMs) and related structures within the general framework of probabilistic independence networks (PINs). The paper presents a self-contained review of the basic principles of PINs. It is shown that the well-known forward-backward (F-B) and Viterbi algorithms for HMMs are special cases of more general inference algorithms for arbitrary PINs. Furthermore, the existence of inference and estimation algorithms for more general graphical models provides a set of analysis tools for HMM practitioners who wish to explore a richer class of HMM structures. Examples of relatively complex models to handle sensor fusion and coarticulation in speech recognition are introduced and treated within the graphical model framework to illustrate the advantages of the general approach.
43
Bayesian Model Selection in Social Research
Adrian E. Raftery · Sociological Methodology · 1995 · 6.9K citations
Computing, Biostatistics, Sociology +17