The simple Bayesian classifier is known to be optimal when attributes are independent given the class, but the question of whether other sufficient conditions for its optimality exist has so far not been explored. Empirical results showing that it performs surprisingly well in many domains containing clear attribute dependences suggest that the answer to this question may be positive. This article shows that, although the Bayesian classifier's probability estimates are only optimal under quadratic loss if the independence assumption holds, the classifier itself can be optimal under zero-one loss (misclassification rate) even when this assumption is violated by a wide margin. The region of quadratic-loss optimality of the Bayesian classifier is in fact a second-order infinitesimal fraction of the region of zero-one optimality. This implies that the Bayesian classifier has a much greater range of applicability than previously thought. For example, in this article it is shown to be opti...

We describe Syskill & Webert, a software agent that learns to rate pages on the World Wide Web (WWW), deciding what pages might interest a user. The user rates explored pages on a three point scale, and Syskill & Webert learns a user profile by analyzing the information on a page. The user profile can be used in two ways. First, it can be used to suggest which links a user would be interested in exploring. Second, it can be used to construct a LYCOS query to find pages that would interest a user. We compare four different learning algorithms and TF-IDF, a standard information retrieval algorithm on this task. . 1 Introduction There is a vast amount of information on the World Wide Web (WWW) and more is becoming available daily. How can a user locate information that might be useful to that user? In this paper, we discuss Syskill & Webert, a software agent that learns a profile of a user's interest, and uses this profile to identify interesting web pages in two ways. First, by having ...

. We discuss algorithms for learning and revising user profiles that can determine which World Wide Web sites on a given topic would be interesting to a user. We describe the use of a naive Bayesian classifier for this task, and demonstrate that it can incrementally learn profiles from user feedback on the interestingness of Web sites. Furthermore, the Bayesian classifier may easily be extended to revise user provided profiles. In an experimental evaluation we compare the Bayesian classifier to computationally more intensive alternatives, and show that it performs at least as well as these approaches throughout a range of different domains. In addition, we empirically analyze the effects of providing the classifier with background knowledge in form of user defined profiles and examine the use of lexical knowledge for feature selection. We find that both approaches can substantially increase the prediction accuracy. Keywords: Information filtering, intelligent agents, multistrategy lea...

In this paper, we examine previous work on the naive Bayesian classifier and review its limitations, which include a sensitivity to correlated features. We respond to this problem by embedding the naive Bayesian induction scheme within an algorithm that carries out a greedy search through the space of features. We hypothesize that this approach will improve asymptotic accuracy in domains that involve correlated features without reducing the rate of learning in ones that do not. We report experimental results on six natural domains, including comparisons with decision-tree induction, that support these hypotheses. In closing, we discuss other approaches to extending naive Bayesian classifiers and outline some directions for future research.

Abstract. Many KDD systems incorporate an implicit or explicit preference for simpler models, but this use of “Occam’s razor ” has been strongly criticized by several authors (e.g., Schaffer, 1993; Webb, 1996). This controversy arises partly because Occam’s razor has been interpreted in two quite different ways. The first interpretation (simplicity is a goal in itself) is essentially correct, but is at heart a preference for more comprehensible models. The second interpretation (simplicity leads to greater accuracy) is much more problematic. A critical review of the theoretical arguments for and against it shows that it is unfounded as a universal principle, and demonstrably false. A review of empirical evidence shows that it also fails as a practical heuristic. This article argues that its continued use in KDD risks causing significant opportunities to be missed, and should therefore be restricted to the comparatively few applications where it is appropriate. The article proposes and reviews the use of domain constraints as an alternative for avoiding overfitting, and examines possible methods for handling the accuracy–comprehensibility trade-off.

Naive Bayesian classifiers which make independence assumptions perform remarkably well on some data sets but poorly on others. We explore ways to improve the Bayesian classifier by searching for dependencies among attributes. We propose and evaluate two algorithms for detecting dependencies among attributes and show that the backward sequential elimination and joining algorithm provides the most improvement over the naive Bayesian classifier. The domains on which the most improvement occurs are those domains on which the naive Bayesian classifier is significantly less accurate than a decision tree learner. This suggests that the attributes used in some common databases are not independent conditioned on the class and that the violations of the independence assumption that affect the accuracy of the classifier can be detected from training data. 23.1 Introduction The Bayesian classifier (Duda

Naive Bayes is an efficient and effective learning algorithm, but previous results show that its representation ability is severely limited since it can only represent certain linearly separable functions in the binary domain. We give necessary and sufficient conditions on linearly separable functions in the binary domain to be learnable by Naive Bayes under uniform representation. We then show that the learnability (and error rates) of Naive Bayes can be affected dramatically by sampling distributions. Our results help us to gain a much deeper understanding of this seemingly simple, yet powerful learning algorithm.

The naive Bayesian classifier provides a simple and e#ective approach to classifier learning, but its attribute independence assumption is often violated in the real world. A number of approaches have sought to alleviate this problem. A Bayesian tree learning algorithm builds a decision tree, and generates a local naive Bayesian classifier at each leaf. The tests leading to a leaf can alleviate attribute inter-dependencies for the local naive Bayesian classifier. However, Bayesian tree learning still su#ers from the small disjunct problem of tree learning. While inferred Bayesian trees demonstrate low average prediction error rates, there is reason to believe that error rates will be higher for those leaves with few training examples. This paper proposes the application of lazy learning techniques to Bayesian tree induction and presents the resulting lazy Bayesian rule learning algorithm, called Lbr. This algorithm can be justified by a variant of Bayes theorem which supports a weaker conditional attribute independence assumption than is required by naive Bayes. For each test example, it builds a most appropriate rule with a local naive Bayesian classifier as its consequent. It is demonstrated that the computational requirements of Lbr are reasonable in a wide cross-section of natural domains. Experiments with these domains show that, on average, this new algorithm obtains lower error rates significantly more often than the reverse in comparison to a naive Bayesian classifier, C4.5, a Bayesian tree learning algorithm, a constructive Bayesian classifier that eliminates attributes and constructs new attributes using Cartesian products of existing nominal attributes, and a lazy decision tree learning algorithm. It also outperforms, although the result is not statisticall...

Predictive accuracy has been used as the main and often only evaluation criterion for the predictive performance of classification learning algorithms. In recent years, the area under the ROC (Receiver Operating Characteristics) curve, or simply AUC, has been proposed as an alternative single-number measure for evaluating learning algorithms. In this paper, we prove that AUC is a better measure than accuracy. More specifically, we present rigourous definitions on consistency and discriminancy in comparing two evaluation measures for learning algorithms. We then present empirical evaluations and a formal proof to establish that AUC is indeed statistically consistent and more discriminating than accuracy. Our result is quite significant since we formally prove that, for the first time, AUC is a better measure than accuracy in the evaluation of learning algorithms.

We describe a software agent that learns to find information on the World Wide Web (WWW), deciding what new pages might interest a user. The agent maintains a separate hotlist (for links that were interesting) and coldlist (for links that were not interesting) for each topic. By analyzing the information immediately accessible from each link, the agent learns the types of information the user is interested in. This can be used to inform the user when a new interesting page becomes available or to order the user's exploration of unseen existing links so that the more promising ones are investigated first. We compare four different learning algorithms on this task. We describe an experiment in which a simple Bayesian classifier acquires a user profile that agrees with a user's judgment over 90% of the time.