Results 1  10
of
13
Learning the structure of dynamic probabilistic networks
, 1998
"... Dynamic probabilistic networks are a compact representation of complex stochastic processes. In this paper we examine how to learn the structure of a DPN from data. We extend structure scoring rules for standard probabilistic networks to the dynamic case, and show how to search for structure when so ..."
Abstract

Cited by 267 (14 self)
 Add to MetaCart
Dynamic probabilistic networks are a compact representation of complex stochastic processes. In this paper we examine how to learn the structure of a DPN from data. We extend structure scoring rules for standard probabilistic networks to the dynamic case, and show how to search for structure when some of the variables are hidden. Finally, we examine two applications where such a technology might be useful: predicting and classifying dynamic behaviors, and learning causal orderings in biological processes. We provide empirical results that demonstrate the applicability of our methods in both domains. 1
Inference and Learning in Hybrid Bayesian Networks
, 1998
"... We survey the literature on methods for inference and learning in Bayesian Networks composed of discrete and continuous nodes, in which the continuous nodes have a multivariate Gaussian distribution, whose mean and variance depends on the values of the discrete nodes. We also briefly consider hybrid ..."
Abstract

Cited by 32 (2 self)
 Add to MetaCart
(Show Context)
We survey the literature on methods for inference and learning in Bayesian Networks composed of discrete and continuous nodes, in which the continuous nodes have a multivariate Gaussian distribution, whose mean and variance depends on the values of the discrete nodes. We also briefly consider hybrid Dynamic Bayesian Networks, an extension of switching Kalman filters. This report is meant to summarize what is known at a sufficient level of detail to enable someone to implement the algorithms, but without dwelling on formalities.
A join tree probability propagation architecture for semantic modeling
 J INTELL INF SYST
, 2008
"... ..."
Probabilistic Reasoning in Bayesian Networks: A Relational Database Approach
 Sixteenth Canadian Conference on Artificial Intelligence
, 2003
"... Probabilistic reasoning in Bayesian networks is normally conducted on a junction tree by repeatedly applying the local propagation whenever new evidence is observed. In this paper, we suggest to treat probabilistic reasoning as database queries. We adapt a method for answering queries in databas ..."
Abstract

Cited by 6 (3 self)
 Add to MetaCart
(Show Context)
Probabilistic reasoning in Bayesian networks is normally conducted on a junction tree by repeatedly applying the local propagation whenever new evidence is observed. In this paper, we suggest to treat probabilistic reasoning as database queries. We adapt a method for answering queries in database theory to the setting of probabilistic reasoning in Bayesian networks. We show an e#ective method for probabilistic reasoning without repeated application of local propagation whenever evidence is observed.
Some Strategies for Explanations in Evidential Reasoning
 IEEE Trans. SMC
, 1994
"... We present two methods to provide explanations for reasoning with belief functions. One approach, inspired by Strat's method, is based on sensitivity analysis, but its computation is simpler thus easier to implement than Strat's. The other approach is to examine the impact of each piece ..."
Abstract

Cited by 6 (1 self)
 Add to MetaCart
We present two methods to provide explanations for reasoning with belief functions. One approach, inspired by Strat's method, is based on sensitivity analysis, but its computation is simpler thus easier to implement than Strat's. The other approach is to examine the impact of each piece of evidence on the conclusion based on the measure of the information content provided by the evidence. We show the property of additivity for each pieces of evidence that are conditional independent within the context of the valuationbased systems. We will give an example to show how these approaches are applied in an evidential network. 1 Introduction The developers of expert systems have realized that a good facility to explain the computerbased reasoning to users is a prerequisite to their more widespread acceptance. The importance of explanation is due to two reasons. First, expert systems are usually used to solve difficult problems. A good explanation facility allows users to observe ...
Generating Explanations for Evidential Reasoning
 In P. Besnard & S. Hanks (Eds.), Proceedings of the Eleventh Conference on Uncertainty in Artificial Intelligence (pp. 574581
, 1995
"... In this paper, we present two methods to provide explanations for reasoning with belief functions in the valuationbased systems. One approach, inspired by Strat's method, is based on sensitivity analysis, but its computation is simpler thus easier to implement than Strat's. The othe ..."
Abstract

Cited by 5 (0 self)
 Add to MetaCart
In this paper, we present two methods to provide explanations for reasoning with belief functions in the valuationbased systems. One approach, inspired by Strat's method, is based on sensitivity analysis, but its computation is simpler thus easier to implement than Strat's. The other one is to examine the impact of evidence on the conclusion based on the measure of the information content in the evidence. We show the property of additivity for the pieces of evidence that are conditional independent within the context of the valuationbased systems. We will give an example to show how these approaches are applied in an evidential network. 1 Introduction The developers of expert systems have realized that a good facility to explain the computerbased reasoning to users is a prerequisite to their more widespread acceptance. The importance of explanation is due to two reasons. First, expert systems are usually used to solve difficult problems. A good explanation facility a...
Local Computation in Covering Join Trees Part #2 Updating in Local Computation ∗
, 2006
"... Local computation on covering join trees provides a solution for query answering in several different fields, such as relational databases, belief functions, constraint satisfaction, Gaussian potentials and many more. The algebraic structure behind is a generic framework for information processing k ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
(Show Context)
Local computation on covering join trees provides a solution for query answering in several different fields, such as relational databases, belief functions, constraint satisfaction, Gaussian potentials and many more. The algebraic structure behind is a generic framework for information processing known as valuation algebras (Kohlas, 2003). In this paper we discuss how new information pieces can be added to the old information, i.e. how to use former computed results to answer the queries in the new situation. A task which will be referred to as updating. The domains of the new information pieces are possibly not covered by any node of the join tree. Since the construction of a new covering join tree may be computationally expensive and makes it harder or even impossible to reuse already available results, we introduce several methods to modify join trees locally. We will see that this enables updating approaches for all wellknown architectures for local compuation like ShenoyShafer, LauritzenSpiegelhalter and HUGIN. As
Updating in Local Computation
"... Abstract. Local computation on covering join trees provides a solution for query answering in several different fields, such as relational databases, belief functions, constraint satisfaction, Gaussian potentials and many more. The algebraic structure behind is a generic framework for information pr ..."
Abstract
 Add to MetaCart
(Show Context)
Abstract. Local computation on covering join trees provides a solution for query answering in several different fields, such as relational databases, belief functions, constraint satisfaction, Gaussian potentials and many more. The algebraic structure behind is a generic framework for information processing known as valuation algebras [5]. In this paper we discuss how new information pieces can be added to the old information, i.e. how to use former computed results to answer the queries in the new situation. A task which will be referred to as updating. The domains of the new information pieces are possibly not covered by any node of the join tree. Since the construction of a new covering join tree may be computationally expensive and makes it harder or even impossible to reuse already available results, we introduce several methods to modify join trees locally. We will see that this enables updating approaches for all wellknown architectures for local compuation like ShenoyShafer, LauritzenSpiegelhalter and HUGIN. 1
Answering Frequent Probabilistic Inference Queries in Databases
, 2011
"... Existing solutions for probabilistic inference queries mainly focus on answering a single inference query, but seldom address the issues of efficiently returning results for a sequence of frequent queries, which is more popular and practical in many real applications. In this paper, we mainly study ..."
Abstract
 Add to MetaCart
Existing solutions for probabilistic inference queries mainly focus on answering a single inference query, but seldom address the issues of efficiently returning results for a sequence of frequent queries, which is more popular and practical in many real applications. In this paper, we mainly study the computation caching and sharing among a sequence of inference queries in databases. The clique tree propagation (CTP) algorithm is first introduced in databases for probabilistic inference queries. We use the materialized views to cache the intermediate results of the previous inference queries, which might be shared with the following queries, and consequently reduce the time cost. Moreover, we take the query workload into account to identify the frequently queried variables. To optimize probabilistic inference queries with CTP, we cache these frequent query variables into the materialized views to maximize the reuse. Due to the existence of different query plans, we present heuristics to estimate costs and select the optimal query plan. Finally, we present the experimental evaluation in relational databases to illustrate the validity and superiority of our approaches in answering frequent probabilistic inference queries.