Intelligent Data Analysis: An International Journal

We proposed a kernel-based binary classification algorithm, named support vector class description (SVCD), which is an extended version of support vector domain description (SVDD) for one-class classification. SVCD constructs two compact hyperspheres in the feature space such that each hypersphere includes as many instances as possible of one class, while keeping the instances of the other class away from the sphere. By doing this, two linearly non-separable classes in the input space can be well distinguished in the feature space. In order to verify the classification performance and exploit the properties of SVCD, we conducted experiments on actual classification data sets and analyzed the results. Compared with other popular kernel-based classification algorithms, such as support vector machine (SVM) and kernel Fisher discriminant analysis (KFD), SVCD gave better classification performances in terms of both the area under the receiving operator curve (AUROC) and the balanced correction rate (BCR). In addition, SVCD was found to be capable of finding moderate sparse solutions with little parameter sensitivity.

A general assumption in all existing algorithms permitting to mine functional dependencies is that the database is static. However, real life databases are frequently updated. To the best of our knowledge, the discovery of functional dependencies in dynamic databases has never been studied. A naïve solution consists in re-applying one of the existing algorithms to discover functional dependencies holding on the updated database. Nevertheless, in many domains, where response time is crucial, re-executing algorithms from scratch would be inacceptable. In this paper, we propose a new technique that makes use of the previously discovered results to cut down the amount of work that has been done to discover the new set of functional dependencies satisfied by the updated database.

It is often necessary to combine multiple scores into a joint decision in many real-world applications. To that end, an essential challenge is to build a proper weighted scoring rule, which assigns more weight to the more important scores and derives a final scoring value. In this paper, we first introduce some desirable properties of weighted scoring rules, and then propose a scheme that can incorporate weights into scores in a natural way. In addition, we show our method has overcome some drawbacks of a classic weighted scoring method. Finally, as a case study, we apply these weighted scoring methods to outlier detection to evaluate how much they help solve the ultimate ranking problem.

This paper is concerned with an application of ICA for a possible improvement of iris recognition by replying to the question: does ICA perform well for such purpose? To achieve this, the hypotheses and the theoretical concepts of ICA methods used are handled so that coherence with iris authentication application is guaranteed. Our contribution is not in the development of new theoretical concepts of ICA but it is in adapting its basic ideas for our application. Also, it consists of deploying its powerful and its efficiency for iris recognition, and consequently; it's potential to embed it on smart cards for increasing application domains of secure biometric-based individual identification systems. We have developed a comparative study between the implemented ICA algorithms and other recent and popular methods of iris recognition. We have demonstrated our experimental results using some mathematical criteria. Three different subsets of international certified CASIA iris image databases are used for testing the different implemented methods. The conclusion of such comparative study is that the ICA-based approaches are more effective and more practical than other existing methods.

Long-term travel time prediction (TTP) can be an important planning tool for both freight transport and public transport companies. In both cases it is expected that the use of long-term TTP can improve the quality of the planned services by reducing the error between the actual and the planned travel times. However, for reasons that we try to stretch out along this paper, long-term TTP is almost not mentioned in the scientific literature. In this paper we discuss the relevance of this study and compare three non-parametric state-of-the-art regression methods: Projection Pursuit Regression (PPR), Support Vector Machine (SVM) and Random Forests (RF). For each one of these methods we study the best combination of input parameters. We also study the impact of different methods for the pre-processing tasks (feature selection, example selection and domain values definition) in the accuracy of those algorithms. We use bus travel time's data from a bus dispatch system. From an off-the-shelf point-of-view, our experiments show that RF is the most promising approach from the three we have tested. However, it is possible to obtain more accurate results using PPR but with extra pre-processing work, namely on example selection and domain values definition.

From the beginning of sequential pattern mining to the present, this field has received important attention within the data mining area, because it has a wide application in several significant computational problems. Many algorithms have been created and several techniques have been used with the objective of improving the discovery of the frequent sequence set. In this paper we present the main characteristics of some of the most important sequential pattern mining algorithms. Also, we show a comparative performance study among these algorithms.

In this paper, a method for telecommunications fraud detection is proposed. The method is based on the user profiling using the Latent Dirichlet Allocation (LDA). Fraudulent behavior is detected with use of a threshold-type classification algorithm, allocating the telecommunication accounts into one of two classes: fraudulent account and non-fraudulent account. The paper provides also a method for automatic threshold computation. The accounts are classified with use of the Kullback-Leibler divergence (KL-divergence). Therefore, we also introduce three methods for approximating the KL-divergence between two LDAs. Finally, the results of experimental study on KL-divergence approximation and fraud detection in telecommunications are reported.

The rapid increase in the amount of textual data has brought forward a growing research interest towards mining text to detect deviations. Specialized methods for specific domains have emerged to satisfy various needs in discovering rare patterns in text. This paper focuses on a graph-based approach for text representation and presents a novel error tolerance dissimilarity algorithm for deviation detection. We resolve two non-trivial problems, i.e. semantic representation of text and the complexity of graph matching. We employ conceptual graphs interchange format (CGIF) – a knowledge representation formalism to capture the structure and semantics of sentences. We propose a novel error tolerance dissimilarity algorithm to detect deviations in the CGIFs. We evaluate our method in the context of analyzing real world financial statements for identifying deviating performance indicators. We show that our method performs better when compared with two related text based graph similarity measuring methods. Our proposed method has managed to identify deviating sentences and it strongly correlates with expert judgments. Furthermore, it offers error tolerance matching of CGIFs and retains a linear complexity with the increasing number of CGIFs.

Ontologies allow us to represent knowledge and data in implicit and explicit ways. Implicit knowledge can be derived by means of several deductive logic-based processes. This paper introduces a new way for extracting implicit knowledge from ontologies by means of a link analysis of the T-box of the ontology integrated with a data mining step on the A-box.

The implicit knowledge extracted is in the form of "Influence Rules" i.e. rules structured as: if property p₁ of concept c₁ has value v₁, then property p₂ of concept c₂ has value v₂ with probability π.

The technique is completely general and applicable to whatever domain. The Influence Rules can be used to integrate existing knowledge or to support any other data mining process.

A case study about an ontology that describes intrusion detection is used to illustrate how the method works.