Instance selection and feature selection are two orthogonal methods for reducing the amount and complexity of data. Feature selection aims at the reduction of redundant features in a dataset whereas instance selection aims at the reduction of the number of instances. So far, these two methods have mostly been considered in isolation. In this paper, we present a new algorithm, which we call FIS (Feature and Instance Selection) that targets both problems simultaneously in the context of text classificationOur experiments on the Reuters and 20-Newsgroups datasets show that FIS considerably reduces both the number of features and the number of instances. The accuracy of a range of classifiers including Naïve Bayes, TAN and LB considerably improves when using the FIS preprocessed datasets, matching and exceeding that of Support Vector Machines, which is currently considered to be one of the best text classification methods. In all cases the results are much better compared to Mutual Information based feature selection. The training and classification speed of all classifiers is also greatly improved.

Nave Bayes (NB) classifier has long been considered a core methodology in text classification mainly due to its simplicity and computational efficiency. There is an increasing need however for methods that can achieve higher classification accuracy while maintaining the ability to process large document collections. In this paper we examine text categorization methods from a perspective that considers the tradeoff between accuracy and scalability to large data sets and large feature sizes. We start from the observation that Support Vector Machines, one of the best text categorization methods cannot scale up to handle the large document collections involved in many real word problems. We then consider bayesian extensions to NB that achieve higher accuracy by relaxing its strong independence assumptions. Our experimental results show that LB, an association-based lazy classifier can achieve a good tradeoff between high classification accuracy and scalability to large document collections...

Itemsets provide local descriptions of the data.This work proposes to use itemsets as basic means for classification purposes too.To enable this, the concept of class support SUpi of an itemset is introduced, i.e., how many times an itemset occurs when a specific class Ci is present.Class supports of frequent itemsets are computed in the training phase.Upon arrival of a new case to be classified, some of the generated itemsets are selected and their class supports SUpi are used to compute the probability that the case belongs to class cc The result is the class ci with highest such probability.We show that selecting and combining many and long itemsets providing new evidence ( interesting) is an effective strategy for computing the class probabilities.The proposed classification technique is called Large Bayes as it happens to reduce to Naive Bayes classifier when all itemsets selected are of size one only.Experimental results on a large number of benchmark data sets show that Large Bayes consistently outperforms the widely used Ndive Bayes classifier.In many cases, Large Bayes is also superior to other state of the art classification methods such as C4.5, CBA (a recently proposed rule classifier built from association rules), and TAN (a Bayesian network extension of Ndive Bayes).

This paper deals with queries involving the retrieval of images that contain certain object configurations. Consider, for instance, that a user wants to "find all images where there exists a building adjacent to the west side of a park which is southwest and near a commercial center". This query can be formulated as a constraint satisfaction problem (CSP) where the query variables are nodes of the corresponding constraint network and the image objects constitute the domain of each variable. The arcs of the network correspond to spatial constraints (e.g., adjacent ^ west (X1,X2), southwest ^ near (X2,X3)). Problems of the above nature are, in general, intractable. In addition, spatial constraints (e.g., southwest, near) lack universally accepted semantics and cannot always be modeled by crisp relations; a fact that further complicates query processing. This paper focuses on the development of effective methods that take advantage of the special structure of the spatial domain to achieve good average performance even for large images and queries.