Ezawa, K., Norton, S. Knowledge discovery in telecommunication services data using Bayesian network models. Proceedings of the First International Conference on Knowledge Discovery and Data Mining; 1995 August 20-21. Montreal Canada. AAAI Press: Menlo Park, CA, 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....The main problem solved here is to generalize the database by the attributes. 2.2.1 Bayesian Network The Bayesian Network is built to show the relationship between the causations and effects. The main idea is to use the Bayesian statistics to build such form of relationship. For example, in [13], Kazuo J. Ezawa et al. introduce the Bayesian Network by using the joint probability Pr( X) of classes and variables X to find the posterior probabilityPr(nX) given the conditional probability of the attributes of the class Pr(Xn) and the unconditional probability of the class Pr( The ....

Kazuo J. Ezawa and Steven W. Norton. Knowledge discovery in telecommunication services data using bayesian network models. In Proceedings of the First International ConferenceonKnowledge Discovery and Data Mining, pages 100--105, 1995.

....distributions may be better suited to capture the dependence structure. To keep the resulting graph sparse, one may introduce the restriction that no attribute may have more than a xed number of parents. Probabilistic networks of this type have been successfully applied in telecommunication [7]. As an illustrative example, let us take a look at the TABLE I A naive Bayes classifier for the iris data. The normal distributions are described by . iris type iris setosa iris versicolor iris virginica prior prob. 0.333 0.333 0.333 petal length 1:46 0:17 4:26 0:46 5:55 ....

K.J. Ezawa and S.W. Norton. Knowledge Discovery in Telecommunication Services Data Using Bayesian Network Models. Proc. 1st Int. Conf. on Knowledge Discovery and Data Mining (KDD'95, Montreal, Canada), 100-105. AAAI Press, Menlo Park, CA, USA 1995

....transactions are fraudulent. In our domain, we found that DC 1 significantly improves detection performance over systems that use transaction classification alone. It would be interesting to determine whether a system like DC 1 could improve performance on these other superimposition fraud tasks. Ezawa and Norton (1995, 1996) have addressed the problem of uncollectible debt in telecommunications services. They use a goal directed Bayesian network for classification, which distinguishes customers who are likely to default from those who are not. As with our work, Ezawa and Norton s work faces problems with ....

Ezawa, K. and S. Norton (1995, August). Knowledge discovery in telecommunication services data using bayesian network models. In U. Fayyad and R. Uthurusamy (Eds.), Proceedings of First International Conference on Knowledge Discovery and Data Mining, Menlo Park, CA, pp. 100--105. AAAI Press.

....to prepare the patterns emerging from the learning module for use. Such postprocessing includes knowledge evaluation, correction and filtering. In recent years, much research related to KDD has been performed. They have been used in aerospace [3] water quality [4] Telecommunication service [5], bank business [6] and textual databases [7] and commercial data warehousing [8, 9] A data warehousing framework is a kind of simple KDD application. One characteristic of such a framework is that the target data and patterns are previously well defined. This greatly restricts application of ....

K. J. Ezawa and S. W. Norton, Knowledge Discovery in Telecommunication Services DAta Using Bayesian Network Models, Proc. of 1st int. conf. on Knowledge Discovery and Data Mining (KDD-95) (1995) 100 -- 105.

....occurred at an accuracy somewhat lower than optimal. In other words, some classification accuracy could be sacrificed to decrease cost. More sophisticated methods could be used to produce cost sensitive classifiers, which would probably produce better results. Related Work Yuhas (1993) and Ezawa and Norton (1995) address the problem of uncollectible debt in telecommunications services. However, neither work deals with characterizing typical customer behavior, so mining the data to derive profiling features is not necessary. Ezawa and Norton s method of evidence combining is much more sophisticated than ....

Ezawa, K., and Norton, S. 1995. Knowledge discovery in telecommunication services data using bayesian network models. In Fayyad, U., and Uthurusamy, R., eds., Proceedings of First International Conference on Knowledge Discovery and Data Mining, 100-- 105. Menlo Park, CA: AAAI Press.

....of learning, one can use this structure to explicitly encode priors, a feature absent in many learning frameworks. Bayesian networks are being increasingly used in various real world applications, including medical diagnosis (Suermondt Amylon, 1989; Andreassen et al. 1987) telecommunications (Ezawa Norton, 1995), information retrieval (Fung Favero, 1995) system troubleshooting (Heckerman et al. 1994a) vision (Levitt et al. 1989) and language understanding (Charniak Goldman, 1989) This paper presents K2 AS, a novel method for inducing selective Bayesian networks, and compares its classification ....

Ezawa, K. & Norton, S. (1995). Knowledge discovery in telecommunication services data using Bayesian network models. In Proc. 1st Int. Conf. on Knowledge Discovery and Data Mining.

.... [Ezawa and Schuermann 1995a,b] Comparison of the performance of several conditionally independent probabilistic models to the performance of conditionally dependent models constructed by APRI using large call detail datasets of 4 6 million records and 600 800 million bytes can be found in [Ezawa and Norton 1995]. 2 THE BAYESIAN NETWORK APPROACH Theoretically, the Bayesian Classifier [Fukunaga 1990] provides optimal classification performance. As a practical matter, however, its implementation is infeasible. Recent advances in evidence propagation algorithms [Shachter 1990, Lauritzen and Spiegelhalter ....

Ezawa, K. J., and Norton S., (1995) Knowledge Discovery in Telecommunication Services Data Using Bayesian Networks. Proceedings of the First International Conference on Knowledge Discovery & Data Mining, Montreal, Canada.

No context found.

Ezawa, K., Norton, S. Knowledge discovery in telecommunication services data using Bayesian network models. Proceedings of the First International Conference on Knowledge Discovery and Data Mining; 1995 August 20-21. Montreal Canada. AAAI Press: Menlo Park, CA, 1995.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC