View Full Document

A NEW OBJECT-ORIENTED STOCHASTIC MODELING LANGUAGE



View the full content.
View Full Document
View Full Document

3 views

Unformatted text preview:

A NEW OBJECT ORIENTED STOCHASTIC MODELING LANGUAGE DAN PLESS GEORGE LUGER Department of Computer Science University of New Mexico Albuquerque NM 87131 ABSTRACT A new language and inference algorithm for stochastic modeling is presented This work refines and generalizes the stochastic functional language originally proposed by 1 The language supports object oriented representation and recursive functions It provides a compact representation for a large class of stochastic models including infinite models It provides the ability to represent general and abstract stochastic relationships and to decompose large models into smaller components Our work extends the language of 1 by providing object encapsulation and reuse and a new and effective strategy for caching An exact and complete inference algorithm is presented here that is expected to support efficient inference over important classes of models and queries KEYWORDS Bayesian Networks Stochastic Modeling Probabilistic Reasoning INTRODUCTION This paper describes a new object oriented stochastic modeling language The language is capable of representing a larger class of models than those expressible as Bayesian Networks It supports a powerful form of object oriented representation allowing general probabilistic relationships over object classes to be expressed and computed The inference algorithm for this language achieves efficiency through modular representation lazy evaluation goal directed inference and compact factoring of conditional probability tables The limitations of flat Bayesian Networks that use simple random variables has been noted by other researchers 2 3 These limitations have motivated a variety of recent research in hierarchical and composable Bayesian models 4 5 6 7 Most of these new Bayesian modeling formalisms support model decomposition often based on an object oriented approach While these provide more expressive and succinct representational frameworks few of these change the class of models



Access the best Study Guides, Lecture Notes and Practice Exams

Loading Unlocking...
Login

Join to view A NEW OBJECT-ORIENTED STOCHASTIC MODELING LANGUAGE and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view A NEW OBJECT-ORIENTED STOCHASTIC MODELING LANGUAGE and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?