ALBERT

Description: Outlier detection is an important task in data mining because outliers may bring either new knowledge or potential threats. Much of recent research has focused on measuring the local difference between an outlier and its nearest neighbors, some of which may be unsuitable reference objects. Thus, local difference cannot represent true outlying-ness. On the basis of this conclusion, we propose a new outlying-ness measure that reflects the connectivity of any object to the main body of a data set. For any object p, the outlying-ness is denoted by the connectivity from the k-th most similar neighbor to p. The proposed measure is applicable to arbitrary-density and arbitrarily-shaped data. It is uninfluenced by unsuitable reference objects and effectively identifies outlying clusters without the need for clustering algorithms and additional parameters. Content Type Journal Article Pages 145-160 DOI 10.3233/IDA-140701 Authors Jiaqiang Wan, Chongqing Key Laboratory of Software Theory and Technology, College of Computer Science, Chongqing University, Chongqing, China Qingsheng Zhu, Chongqing Key Laboratory of Software Theory and Technology, College of Computer Science, Chongqing University, Chongqing, China Dajiang Lei, College of Computer, Chongqing University of Posts and Telecommunications, Chongqing, China Jiaxi Lu, People's Procuratorate of YuBei District, Chongqing, China Journal Intelligent Data Analysis Online ISSN 1571-4128 Print ISSN 1088-467X Journal Volume Volume 19 Journal Issue Volume 19, Number 1 / 2015

Description: Intrusion Detection Systems (IDS) are necessary and important tools for monitoring information systems. However they produce a huge quantity of alerts. Alerts correlation is a process that reduces the number of alerts reported by intrusion detection systems. In this paper, we propose a new algorithm for a logical-based alerts correlation approach that integrates: security operator's knowledge and preferences. The representation and the reasoning on these knowledge and preferences are done using a new logic called Instantiated First Order Qualitative Choice Logic (IFO-QCL). Our modeling views an alert as an interpretation which allows us to have an efficient algorithm that performs the correlation process in a polynomial time. This paper also provides experimental results which are achieved on datasets issued from a real monitoring system. Content Type Journal Article Pages 3-27 DOI 10.3233/IDA-140693 Authors Lydia Bouzar-Benlabiod, LCSI Laboratory, Ecole Nationale Supérieure d'Informatique, Algiers, Algeria Salem Benferhat, CRIL-CNRS, Université d'Artois, Lens, France Thouraya Bouabana-Tebibel, LCSI Laboratory, Ecole Nationale Supérieure d'Informatique, Algiers, Algeria Journal Intelligent Data Analysis Online ISSN 1571-4128 Print ISSN 1088-467X Journal Volume Volume 19 Journal Issue Volume 19, Number 1 / 2015

Description: Recently, Indicator-based Evolutionary Algorithms are considered the main issue for researchers in the evolutionary multi-objective frameworks. Due to the capability of the Indicator-based approaches in obtaining a finest non-dominated solutions and the potential of these approaches on achieving the well-distributed solutions, these approaches become popular among modern Multi-Objective Evolutionary Algorithms (MOEAs). Most modern MOEAs are intended to converge to the Pareto optimal front through preserving the population diversity in the objective space. In this regard, the intention of this work is presenting a novel MOEA to enhance the population diversity among the non-dominated vectors in the solution space. The idea of this method is inspired by the Hierarchical clustering. In this attitude, an adept approach is planned to present a new indicator as a selection method during the optimization cycle. The gain of this technique is a desirable set with more diverse solutions in the solution space during the environmental selection operator. In the last part, this work also improved the rate of the convergence by introducing a parent selection mechanism. The selection method is simple and effective, which is worked base on the selection of proper members of parents' population instead of a random mechanism. This bright parent selection is adopted to accelerate the convergence of the proposed method. This work is applied to a wide range of well-established test problems. The obtained results validate the motivation on the basis of diversity and performance measures in comparison with the state of the art algorithms. Content Type Journal Article Pages 187-208 DOI 10.3233/IDA-140703 Authors Kamyab Tahernezhad, CSE and IT Department, Shiraz University, Shiraz, Iran Kimia Bazargan Lari, CSE and IT Department, Shiraz University, Shiraz, Iran Ali Hamzeh, CSE and IT Department, Shiraz University, Shiraz, Iran Sattar Hashemi, CSE and IT Department, Shiraz University, Shiraz, Iran Journal Intelligent Data Analysis Online ISSN 1571-4128 Print ISSN 1088-467X Journal Volume Volume 19 Journal Issue Volume 19, Number 1 / 2015

Description: Ensemble is a recently emerged computing technique to provide promising decisions by a consensus of multiple classifiers. The benefit of classifier ensembles has been demonstrated in a vast number of studies in the scope of credit risk management. Yet the performance of different ensemble models was rarely compared when the costs of misclassification errors are asymmetric. In this paper, we concentrate on the performance of 6 ensemble techniques in the context of cost-sensitive credit scoring using 3 financial data sets. The ensemble models are built on the basis of a set of component classifiers derived from different subsets of instances or features by a single learning algorithm. The performance of classifiers is evaluated in terms of expected misclassification cost and compared by nonparametric significance test. The experimental results demonstrate that the functionality of ensembles for boosting the performance of individual classifiers is closely related to the underlying learning algorithms and the employed ensemble techniques. Content Type Journal Article Pages 127-144 DOI 10.3233/IDA-140700 Authors Ning Chen, Instituto Superior de Engenharia do Porto, Porto, Portugal Bernardete Ribeiro, Department of Informatics Engineering, University of Coimbra, Coimbra, Portugal An Chen, Instituto Superior de Engenharia do Porto, Porto, Portugal Journal Intelligent Data Analysis Online ISSN 1571-4128 Print ISSN 1088-467X Journal Volume Volume 19 Journal Issue Volume 19, Number 1 / 2015

Description: One of the prevalent learning tasks involving images is content-based image classification. This is a difficult task especially because the low-level features used to digitally describe images usually capture little information about the semantics of the images. In this paper, we tackle this difficulty by enriching the semantic content of the image representation by using external knowledge. The underlying hypothesis of our work is that creating a more semantically rich representation for images would yield higher machine learning performances, without the need to modify the learning algorithms themselves. The external semantic information is presented under the form of non-positional image labels, therefore positioning our work in a weakly supervised context. Two approaches are proposed: the first one leverages the labels into the visual vocabulary construction algorithm, the result being dedicated visual vocabularies. The second approach adds a filtering phase as a pre-processing of the vocabulary construction. Known positive and known negative sets are constructed and features that are unlikely to be associated with the objects denoted by the labels are filtered. We apply our proposition to the task of content-based image classification and we show that semantically enriching the image representation yields higher classification performances than the baseline representation. Content Type Journal Article Pages 161-185 DOI 10.3233/IDA-140702 Authors Marian-Andrei Rizoiu, ERIC Laboratory, University Lumière Lyon, Avenue Pierre Mendés-France, Bron Cedex, France Julien Velcin, ERIC Laboratory, University Lumière Lyon, Avenue Pierre Mendés-France, Bron Cedex, France Stéphane Lallich, ERIC Laboratory, University Lumière Lyon, Avenue Pierre Mendés-France, Bron Cedex, France Journal Intelligent Data Analysis Online ISSN 1571-4128 Print ISSN 1088-467X Journal Volume Volume 19 Journal Issue Volume 19, Number 1 / 2015

Description: Social tagging provides an effective way for users to organize, manage, share and search for various kinds of resources. These tagging systems have resulted in more and more users providing an increasing amount of information about themselves that could be exploited for recommendation purposes. However, since social tags are generated by users in an uncontrolled way, they can be noisy and unreliable and thus exploiting them for recommendation is a non-trivial task. In this article,a new recommender system is proposed based on the similarities between user and item profiles. The approach here is to generate user and item profiles by discovering frequent user-generated tag patterns. We present a method for finding the underlying meanings (concepts) of the tags, mapping them to semantic entities belonging to external knowledge bases, namely WordNet and Wikipedia, through the exploitation of ontologies created within the W3C Linking Open Data initiative. In this way, the tag-base profiles are upgraded to semantic profiles by replacing tags with the corresponding ontology concepts. In addition, we further improve the semantic profiles through enriching them with a semantic spreading mechanism. To evaluate the performance of this proposed approach, a real dataset from The Del.icio.us website is used for empirical experiment. Experimental results demonstrate that the proposed approach provides a better representation of user interests and achieves better recommendation results in terms of precision and ranking accuracy as compared to existing methods. We further investigate the recommendation performance of the proposed approach in face of the cold start problem and the result confirms that the proposed approach can indeed be a remedy for the problem of cold start users and hence improving the quality of recommendations. Content Type Journal Article Pages 109-126 DOI 10.3233/IDA-140699 Authors Hamed Movahedian, Department of Computer Engineering, Faculty of Engineering, University of Isfahan, Isfahan, Iran Mohammad Reza Khayyambashi, Department of Computer Engineering, Faculty of Engineering, University of Isfahan, Isfahan, Iran Journal Intelligent Data Analysis Online ISSN 1571-4128 Print ISSN 1088-467X Journal Volume Volume 19 Journal Issue Volume 19, Number 1 / 2015

Description: Prediction of missing or potential links and edges is currently the central theme in network analysis. Most of the work is focused on large unlabelled networks, with techniques based on global network models and, on a local level, on using patterns of temporal evolution. We define a problem of small network completion, which deals with sets of small networks, possibly with no recorded temporal dynamics. This problem requires a different set of methods and evaluation procedures. We present a method named Hyspan that extracts frequent patterns from small networks and uses them to predict missing vertices and edges in new networks. It ranks the predicted vertices and edges according to their likelihood estimated from the number and support of the patterns that suggest a particular missing part. Empirical evaluation on real and synthetic data sets shows that the method performs reasonably well. The quality of results depends upon the number and size of the used patterns; a larger number of patterns yields better results but requires longer – although still acceptable – running times. Content Type Journal Article Pages 89-108 DOI 10.3233/IDA-140698 Authors Matija Polajnar, Faculty of Computer and Information Science, University of Ljubljana, Tržaška, Ljubljana, Slovenia Janez Demšar, Faculty of Computer and Information Science, University of Ljubljana, Tržaška, Ljubljana, Slovenia Journal Intelligent Data Analysis Online ISSN 1571-4128 Print ISSN 1088-467X Journal Volume Volume 19 Journal Issue Volume 19, Number 1 / 2015

Description: The bag of words approach describes an image as a histogram of visual words. Therefore, the structural relation between words is lost. Since graphs are well adapted to represent these structural relations, we propose, in this paper, an image classification framework which draws benefit from the efficiency of the graph in modeling structural information and the good classification performances given by the bag of words method. For each image in the dataset, a graph is created by modeling the spatial relations between dense local patches. Thus, we obtain a graph dataset. From the graph dataset, we select the most frequent subgraphs to construct the bag of subgraphs (BoSG) and we associate to each image a subgraph histogram that describes its visual content. For experiments, we have used the two challenging datasets: 15 Scenes and Pascal VOC 2007. Experimental results show that the proposed method outperforms the bag of words and the spatial pyramid models in terms of recognition rate. Content Type Journal Article Pages 75-88 DOI 10.3233/IDA-140697 Authors Mahmoud Mejdoub, Research Groups on Intelligent Machines, University of Sfax, ENIS, Sfax, Tunisia Najib Ben Aoun, Research Groups on Intelligent Machines, University of Sfax, ENIS, Sfax, Tunisia Chokri Ben Amar, Research Groups on Intelligent Machines, University of Sfax, ENIS, Sfax, Tunisia Journal Intelligent Data Analysis Online ISSN 1571-4128 Print ISSN 1088-467X Journal Volume Volume 19 Journal Issue Volume 19, Number 1 / 2015

Description: Association-rule mining is used to mine the relationships among the occurrences itemsets in a transactional database. An item is treated as a binary variable whose value is one if it appears in a transaction and zero otherwise. In real-world applications, several products may be purchased at the same time, with each product having an associated profit, quantity, and price. Association-rule mining from a binary database is thus not sufficient in some applications. Utility mining was thus proposed as an extension of frequent-itemset mining for considering various factors from the user. Most utility mining approaches can only process static databases and use batch processing. In real-world applications, transactions are dynamically inserted into or deleted from databases. The Fast UPdated (FUP) algorithm and the FUP2 algorithm were respectively proposed to handle transaction insertion and deletion in dynamic databases. In this paper, a fast-updated high-utility itemsets for transaction deletion (FUP-HUI-DEL) algorithm is proposed to handle transaction deletion for efficiently updating discovered high utility itemsets in decremental mining. The two-phase approach in high utility mining is applied to the proposed FUP-HUI-DEL algorithm for preserving the downward closure property to reduce the number of candidates. The FUP2 algorithm for handling transaction deletion in association-rule mining is adopted in the proposed FUP-HUI-DEL algorithm to reduce the number of scans of the original database in high utility mining. Experiments show that the proposed FUP-HUI-DEL algorithm outperforms the batch two-phase approach. Content Type Journal Article Pages 43-55 DOI 10.3233/IDA-140695 Authors Chun-Wei Lin, Innovative Information Industry Research Center, School of Computer Science and Technology, Harbin Institute of Technology Shenzhen Graduate School, HIT Campus Shenzhen University Town, Shenzhen, Guangdong, China Guo-Cheng Lan, Computational Intelligence Technology Center, Industrial Technology Research Institute, Hsinchu, Taiwan Tzung-Pei Hong, Department of Computer Science and Information Engineering, National University of Kaohsiung, Kaohsiung, Taiwan Journal Intelligent Data Analysis Online ISSN 1571-4128 Print ISSN 1088-467X Journal Volume Volume 19 Journal Issue Volume 19, Number 1 / 2015

Description: The estimate of the probability density function or probability mass function of an unknown stochastic process is a very important preliminary step for any further elaboration. Most of the traditional approaches to this problem perform a preliminary choice of a parametric mathematical model of the function to estimate and a subsequent fitting on its parameters. To this aim some a-priori knowledge and/or assumptions on the phenomenon under consideration are needed. In this paper an alternative approach is proposed, which does not require any assumption on the available data, as it extracts the probability density function from the output of a neural network, that is trained with a suitable database including the original data and some ad hoc created data with known distribution. The results of the tests performed on synthetic and industrial databases are described and discussed in the paper. Content Type Journal Article Pages 29-41 DOI 10.3233/IDA-140694 Authors Valentina Colla, Electronic Department, Politecnico di Torino, Torino, Italy Marco Vannucci, PERCRO Laboratory, Via L. Alamanni, TeCIP Institute, Scuola Superiore Sant'Anna, Ghezzano, Italy Leonardo M. Reyneri, Electronic Department, Corso Duca Degli Abruzzi, Politecnico di Torino, Torino, Italy Journal Intelligent Data Analysis Online ISSN 1571-4128 Print ISSN 1088-467X Journal Volume Volume 19 Journal Issue Volume 19, Number 1 / 2015