Search

Accurate cancer classification is very important to cancer diagnosis and treatment. As molecular information is increasing for the cancer classification, a lot of techniques have been proposed and utilized to classify and predict the cancers from gene expression profiles. In this paper, we propose a method based on speciated evolution for the cancer classification. The optimal combination among several feature-classifier pairs from the various features and classifiers is evolutionarily searched using the deterministic crowding genetic algorithm. Experimental results demonstrate that the proposed method is more effective than the standard genetic algorithm and the fitness sharing genetic algorithm as well as the best single classifier to search the optimal ensembles for the cancer classification. [ABSTRACT FROM AUTHOR]

This paper suggests an automated method for document classification using an ontology, which expresses terminology information and vocabulary contained in Web documents by way of a hierarchical structure. Ontologybased document classification involves determining document features that represent the Web documents most accurately, and classifying them into the most appropriate categories after analyzing their contents by using at least two pre-defined categories per given document features. In this paper, Web documents are classified in real time not with experimental data or a learning process, but by similar calculations between the terminology information extracted from Web texts and ontology categories. This results in a more accurate document classification since the meanings and relationships unique to each document are determined. Keywords: Document classification, Ontology, Web Page classification. [ABSTRACT FROM AUTHOR]

This paper describes a novel scheme for automatic identification of a species from its genomic data. Random samples of a given length (10,000 elements) are taken from a genome sequence of a particular species. A set of 64 keywords is generated using all possible 3-tuple combinations of the 4 letters: A (for Adenine), T (for Thymine), C (for Cytosine) and G (for Guanine) representing the four types of nucleotide bases in a DNA strand. These 43 = 64 keywords are searched in a sample of the genome sequence and their corresponding frequencies of occurrence are determined. Upon repeating this process for N randomly selected samples taken from the genome sequence, an N × 64 matrix of frequency count data is obtained. Then Principal Component Analysis is employed on this data to obtain a Feature Descriptor of reduced dimension (1 × 64). On determining the feature descriptors of different species and also by taking different samples from the same species, it is found that they are unique for a particular species while wide differences exist between those of different species. The variance of the descriptors for a given genome sequence being negligible, the proposed scheme finds extensive applications in automatic species identification. [ABSTRACT FROM AUTHOR]

This paper presents results of research related to the second stage of the project dedicated to developing and employing the xOMNI system (software of the SCADA type). This stage is associated with the elaboration of the Energy Management System (EMS) for the subtransmission system of the Minas Gerais State Energy Company. One of the most important EMS functions is voltage and reactive power control. Its implementation is based on the use of fuzzy logic technology. Taking this into account, the paper covers the consideration of questions of building diverse types of sensitivity indices and forming production rules providing comprehensive and flexible solutions in conditions of regulated and deregulated environments. Some aspects of rational carrying out fuzzy logic procedures are considered as well. [ABSTRACT FROM AUTHOR]

This paper introduces approaches for power transformer thermal modeling based on two conceptually different recurrent neural networks. The first is the Elman recurrent neural network model whereas the second is a recurrent neural fuzzy network constructed with fuzzy neurons based on triangular norms. These two models are used to model the thermal behavior of power transformers using data reported in literature. The paper details the neural modeling approaches and discusses their main capabilities and properties. Comparisons with the classic deterministic model and static neural modeling approaches are also reported. Computational experiments suggest that the recurrent neural fuzzy-based modeling approach outperforms the remaining models from both, computational processing speed and robustness point of view. [ABSTRACT FROM AUTHOR]

This paper describes an anomaly detection algorithm by monitoring the spin-down of jet engines. The speed profile from multiple engines in an aircraft is cast as a singular value monitoring problem. This relatively simple algorithm is an excellent example of onboard, lightweight feature extractor, the results of which can feed more elaborate trouble shooting procedures. The effectiveness of the algorithm is demonstrated using aircraft data from one of Honeywell's airline customers. [ABSTRACT FROM AUTHOR]

Nominal feature is one type of symbolic features, whose feature values are completely unordered. The most often used existing similarity metrics for symbolic features is the Hamming metric, where similarity computation is coarse-grained and may affect the performance of case retrieval and then the classification accuracy. This paper presents a GA-based approach for learning similarity measure of nominal features for CBR classifiers. Based on the learned similarities, the classification accuracy can be improved, and the importance of each nominal feature can be analyzed to enhance the understanding of the used data sets. [ABSTRACT FROM AUTHOR]

Rough Set Theory is a mathematical tool to deal with vagueness and uncertainty. Rough Set Theory uses a single information table. Relational Learning is the learning from multiple relations or tables. This paper presents a new approach to the extension of Rough Set Theory to multiple relations or tables. The utility of this approach is shown in classification experiments in predictive toxicology. [ABSTRACT FROM AUTHOR]

In this paper we propose a method to identify significant attributes which aid in good classification, as well as introduce certain degrees of roughness. Exception rules are formed with these attributes using the fact that exceptional elements have high rough memberships to more than one distinct class. [ABSTRACT FROM AUTHOR]

Recently rough cluster algorithm were introduced and successfully applied to real life data. In this paper we analyze the rough k-means introduced by Lingras' et al. with respect to its compliance to the classical k-means, the numerical stability and its performance in the presence of outliers. We suggest a variation of the algorithm that shows improved results in these circumstances. [ABSTRACT FROM AUTHOR]

The rough sets theory has proved to be useful in knowledge discovery from databases, decision-making contexts and pattern recognition. However this technique has some difficulties with complex data due to its lack of flexibility and excessive dependency on the initial discretization of the continuous attributes. This paper presents the divisible rough sets as a new hybrid technique of automatic learning able to overcome the problems mentioned using a combination of variable precision rough sets with self-organizing maps and perceptrons. This new technique divides some of the equivalence classes generated by the rough sets method in order to obtain new certain rules under the data which originally were lost. The results obtained demonstrate that this new algorithm obtains a higher decision-making success rate in addition to a higher number of classified examples in the tested data sets. [ABSTRACT FROM AUTHOR]

In this paper, we propose a data mining technique for finding frequently used web pages. These pages may be kept in a server's cache to speed up web access. Existing techniques of selecting pages to be cached do not capture a user's surfing patterns correctly. We use a Weighted Association Rule (WAR) mining technique that finds pages of the user's current interest and cache them to give faster net access. This approach captures both user's habit and interest as compared to other approaches where emphasis is only on habit. Keywords: Data Mining, Web Server, Weighted Association Rule (WAR). [ABSTRACT FROM AUTHOR]

In this paper we have proposed a system that performs both ontology-based text information extraction and ontology update using the extracted information. The system employs text-mining techniques to mine information from text documents guided by an underlying ontology. It also enhances the existing ontology with new concepts and their descriptors which may be precise and/ or imprecise, mined from the text. All extracted information related to concepts and concept descriptors are also stored in a structured knowledge base. Keywords: Text mining, Information extraction, Ontology enhancement. [ABSTRACT FROM AUTHOR]

In this paper, an efficient Hybrid Hierarchical Agglomerative Clustering (HHAC) technique is proposed for effective clustering and prototype selection for pattern classification. It uses the characteristics of both partitional (an incremental scheme) and Hierarchical Agglomerative Clustering (HAC) schemes. Initially, an incremental, partitional clustering algorithm - leader is used for finding the subgroups/subclusters. It reduces the time and space requirements incurred in the formation of the subclusters using the conventional hierarchical agglomerative schemes or other methods. Further, only the subcluster representatives are merged to get a required number of clusters using a hierarchical agglomerative scheme which now requires less space and time when compared to that of using it on the entire training set. Thus, this hybrid scheme would be suitable for clustering large data sets and we can get a hierarchical structure consisting of clusters and subclusters. The subcluster representatives of a cluster can also handle its arbitrary/non-spherical shape. The experimental results (Classification Accuracy (CA) using the prototypes obtained and the computation time) of the proposed algorithm are promising. [ABSTRACT FROM AUTHOR]

View materialization or pre-computation of aggregates(views) is a well known technique used in data warehouse design and Decision Support System(DSS) to reduce the query response time. Obviously, all the views cannot be materialized due to space-time constraint. So, one important decision in designing Data Warehouse and DSS is to select the views to be materialized, which will reduce the query response time to the minimum limit in a DSS . This paper presents a density-based view materialization algorithm with average runtime complexity O(nlogn), where n is the number of views. We have used data cube lattice, view size, access frequency of the views and support(frequency) of the views in selecting the views to be materialized. Our algorithm works much faster and selects better views than other existing algorithms. [ABSTRACT FROM AUTHOR]

In this paper we have proposed a scheme for parsing News video sequences into their semantic components using integrated aural and visual features. We have explored use of the Token Passing Algorithm with HMM for simultaneous segmentation and characterization of the components. Experimentation with about 100 sequences have shown impressive results. [ABSTRACT FROM AUTHOR]

This paper present a novel framework for formal specification of spatio-temporal relations between media objects using fuzzy membership. We have illustrated its use in multimedia ontologies and have described a reasoning framework for creating media based descriptions of concepts. [ABSTRACT FROM AUTHOR]

This paper presents an efficient image retrieval technique based on content using segmentation approach and by considering global distribution of color. To cope with significant appearance changes, the method uses a global size and shape histogram to represent the image regions obtained after segmenting the image based on color similarity. The indexing technique can be found to be significant in comparison to its other counterparts, such as moment based method [12], due to its transformation invariance and effective retrieval performance over several application domains. [ABSTRACT FROM AUTHOR]

Binary images (like cartoons, text documents, signatures captured by signing pads and/or 2-color imagery) are very commonly used in our daily life. Changing the pixel values in these images for hiding the data, may produce a noticeable change in the cover media. The primary problem is the capacity of the embedding technique and preserving the visible artifacts of the cover image even after embedding the secured data. In this paper, we present a run length based steganography algorithm for embedding the secured data into binary images. The proposed algorithm alters pixels of the embeddable blocks of cover image depending on there run length characteristics and chactersitics values of the block. In addition, the new algorithm is based on variable embedding rate for each block, which enhances the security of embedded data and the capacity of the embedding method. We test the performance of the algorithm over 50 various sizes of binary cover images embedding various sizes of secured data. Comparisons with existing algorithms will be presented. Keyword: Run length, steganography, binary image. [ABSTRACT FROM AUTHOR]

In this paper we present a paradigm for robust information hiding strategy designed with proper cryptographic and error correcting frameworks in mind that works for low end hardware devices like mobile phones with camera. Keywords: Steganography, Digital Watermarking, Information Security, JPEG Compression, Mobile Devices. [ABSTRACT FROM AUTHOR]

In this paper, we propose the chromatic adaptation model (CAM) for the Variations of the Luminance Levels in the same Chromaticity Viewing Conditions. The proposed model is obtained by the transform the test colors of the high luminance into the corresponding colors of the low luminance. In the proposed model, the optimal coefficients are obtained from the corresponding colors data of the Breneman's experiments. In the experimental results, we confirmed that the chromaticity errors between the predicted colors by the proposed model and the corresponding colors of the Breneman's experiments are 0.004 in u ′v ′ chromaticity coordinates. The prediction performance of the proposed model is excellent because this error is the threshold value that two adjacent color patches can be distinguished. [ABSTRACT FROM AUTHOR]

Image segmentation techniques are very useful in the analysis of aerial images, biomedical images and seismic images as well as in the automation of industrial applications. The paper presents a few texture-specific features for the segmentation of texture images. We will present four approaches to texture segmentation, viz, descriptor based, heuristic function based, defuzzified feature based and Mask feature based. We have mainly used FCM for texture segmentation. Keywords: Texture segmentation, entropy, descriptors, texture energy measure, defuzzified feature. [ABSTRACT FROM AUTHOR]

In this paper we propose an efficient new algorithm for making an intelligent system for isolating the lung cancer from the inflamed region using needle biopsies. The best way among the cancer treatments, surgery, is the way that can be used for the removal of a malignant tumor in an operation. It is most effective when a cancer is small and localized. Identification and removal of the cancer cells in their earliest formation are very much important. Almost all of the diagnostic laboratories in the world use experts to identify the suspected cells of the lung tumors under microscope. Due to the smaller number of experts, the proposed method, derived based on image contour analysis, has an important significance to replace the manual methods by an intelligent system. [ABSTRACT FROM AUTHOR]

Facial asymmetry has now been established as a useful biometric for human identification in the presence of expression variations ([1]). The current paper investigates an alternative representation of asymmetry in the frequency domain framework, and its significance in identification tasks in terms of the phase component of the frequency spectrum of an image. The importance of the latter in face reconstruction is well-known in the engineering literature ([2]) and this establishes a firm ground for the success of asymmetry as a potentially useful biometric. We also point out some useful implications of this connection and dual representation. Moreover, the frequency domain features are shown to be more robust to intra-personal distortions than the corresponding spatial measures and yield error rates as low as 4% on a dataset with images showing extreme expression variations. [ABSTRACT FROM AUTHOR]

In this paper, a simple and a robust algorithm to detect edges in a gray image is proposed. The statistical property of the small eigenvalue of the covariance matrix of a set of connected pixels over a small region of support is explored for the purpose of edge detection. The gray image is scanned from the top left corner to the bottom right corner with a moving mask of size k xk, for some integer k. At every stage, the small eigenvalue of the covariance matrix of the connected pixels that are having approximately same intensity as that of the center pixel of the mask is computed. This small eigenvalue is used to decide if a pixel is a potential edge pixel based on a pre-defined threshold value. The set of all identified potential edge pixels are then subjected to a pruning process where true edge pixels are selected. Experiments have been conducted on benchmark gray images to establish the performance of the proposed model. Comparative analysis with the Canny edge detector [1] and Sun et al. [15] is made to demonstrate the implementation simplicity and suitability of the proposed method in vision applications. Keywords: Gray image; Small eigenvalue; Eigen image; Edge detection. [ABSTRACT FROM AUTHOR]

Existing shape description techniques using Bezier curves do not adequately consider the domain specific shape information such as the cornerity or gradualness of a shape in the control point generation process. This can lead to large distortion in shape representation even when a large descriptor is used. This paper addresses the issue by introducing a novel improved shape descriptor using Bezier curves (ISDBC) algorithm which divides a shape into segments depending on the cornerity and generates the control points for the segments based on shape information. It also provides an efficient control point encoding strategy which exploits the inherent periodic nature of the distances between consecutive control points. The performance of the ISDBC algorithm has been rigorously tested upon a number of arbitrary shapes, with both quantitative and qualitative results confirming its superiority over existing algorithms. [ABSTRACT FROM AUTHOR]

Among X-ray cross-sectional imaging methods, digital tomosynthesis (DT) is very useful to PCB inspection because it can obtain a cross-sectional image of a local inspection area quickly. The image intensifier, which is usually used in DT systems, distorts X-ray images in shape and intensity. Therefore, image distortion correction is one of the most important issues in realizing a DT system. This paper presents an image distortion correction method to acquire an arbitrary cross-sectional image of an object by using a distance ratio function in an X-ray DT system. The method uses a simplified distortion model made by a distance ratio function in intensity correction, and by the 2D point mapping polynomials in shape correction. [ABSTRACT FROM AUTHOR]

Huge amounts of data are generated in every field of science and technology and the need for the proper data analysis tools and their adaptation to the ever-increasing data size is more and more crucial. Statistical exploratary data analysis techniques -such as principal component analysis, correspondence analysis, clustering and classification among others- are greatly useful in discovering useful information -or knowledge- hidden in data but they require the data set to be complete. In many situations the data is incomplete for various reasons. Erroneous and uncertain data may also be considered as missing since their use may lead to incorrect results. Many research works have addressed this issue in specific applications. This paper presents a simple and efficient iterative method for estimating the missing values in the data set based on linear factorial smoothing. Though this work was prompted by the recurrent problem faced in the field of bioinformatics while analysing the gene expression data, the method proposed for missing value imputation in this paper may be useful in any area. [ABSTRACT FROM AUTHOR]

The configuration of a computational intelligence (CI) method is responsible for its intelligence (e.g. tolerance, flexibility) as well as its accuracy. In this paper, we investigate how to automatically improve the performance of a CI method by finding alternate configuration parameter values that produce more accurate results. We explore this by using a genetic algorithm (GA) to find suitable configurations for the CI methods in an integrated CI system, given several different input data sets. This paper describes the implementation and validation of our approach in the domain of software testing, but ultimately we believe it can be applied in many situations where a CI method must produce accurate results for a wide variety of problems. [ABSTRACT FROM AUTHOR]

The paper investigates a novel technique of designing 2-dimensional IIR digital filters using a modified version of Differential Evolution (DE) where the scalar factor used for weighing the difference vector is made to vary randomly. This approach makes the classical DE more stochastic and provides it with additional exploration capability over the search space. The task of the design has been reformulated as a constrained minimization problem and is solved by the convergence of the proposed algorithm. Numerical example has been provided in support of the theoretical results. The paper also attempts to demonstrate the superiority of the proposed design method by comparing with a few previous design methods based on GA and neural networks. [ABSTRACT FROM AUTHOR]

In this paper, we propose a fuzzy extension to proximal support vector classification via generalized eigenvalues. Here, a fuzzy membership value is assigned to each pattern, and points are classified by assigning them to the nearest of two non parallel planes that are close to their respective classes. The algorithm is simple as the solution requires solving a generalized eigenvalue problem as compared to SVMs, where the classifier is obtained by solving a quadratic programming problem. The approach can be used to obtain an improved classification when one has an estimate of the fuzziness of samples in either class. Keywords: Support vector machines, fuzzy data classification, machine learning, generalized eigenvalue problem, proximal classifier. [ABSTRACT FROM AUTHOR]

To date, image segmentation is used as a vital step to developing handwritten character recognition systems. In the paper, image segmentation has been avoided by applying image fusion and fuzzy method to recognize handwritten characters. At first, several digital images corresponding to each handwritten character are fused to generate patterns, which are stored in 10×10 matrices, irrespective of the size of the images. The character to be recognized is matched with each pattern and the best match pattern is considered as the most likely choice. In the second phase of the work subjective judgment using fuzzy logic has been incorporated to improve the recognition rate where Gaussian distribution function, with varied mean and standard deviation are considered as fuzzy membership functions. Several binary images have been tested to demonstrate the effectiveness of the proposed system. [ABSTRACT FROM AUTHOR]

In this paper, a new adaptive fuzzy controller is proposed for trajectory tracking of wheeled mobile robots by visual servoing. The control algorithm is developed so that it can take care of parametric uncertainty associated with the vision system and the mobile robot dynamics. The system uncertainty associated with nonlinear robot dynamics is estimated by an adaptive fuzzy logic system (FLS) and the uncertain camera parameters are updated online. The controller is designed based on Lyapunov stability theory. Simulation results are presented to illustrate the performance of the proposed controller. [ABSTRACT FROM AUTHOR]

In this paper, we propose a radial basis function (RBF) neural network for detecting a known signal in the presence of non-Gaussian and Gaussian noise. In case of non-Gaussian noise, our study shows that RBF signal detector has significant improvement in performance characteristics; detection capability is better to those obtained with multilayer perceptrons (MLP) and the matched filter (MF) detector. [ABSTRACT FROM AUTHOR]

Functional Magnetic Resonance Imaging (fMRI) allows the neuroscientists to observe the human brain in vivo. The current approach consists in statistically validating their hypotheses. Data mining techniques provide an opportunity to help them in making up their hypotheses. This paper shows how a neuronal clustering technique can highlight active areas thanks to an appropriate distance between fMRI image sequences. This approach has been integrated into an interactive environment for knowledge discovery in brain fMRI. Its results on a typical dataset validate the approach and open further developments in this direction. [ABSTRACT FROM AUTHOR]

A face recognition technique based on symbolic PCA approach is presented in this paper. The proposed method transforms the face images by extracting knowledge from training samples into symbolic objects, termed as symbolic faces. Symbolic PCA is employed to compute a set of subspace basis vectors for symbolic faces and then to project the symbolic faces into the compressed subspace. New test images are then matched with the images in the database by projecting them onto the basis vectors and finding the nearest symbolic face in the subspace. The feasibility of the new symbolic PCA method has been successfully tested for face recognition using ORL database. As compared to eigenface method, the proposed method requires less number of features to achieve the same recognition rate. [ABSTRACT FROM AUTHOR]

The problem of on-line recognition and retrieval of relatively weak industrial signal such as Partial Discharges (PD), buried in excessive noise has been addressed in this paper. The major bottleneck being the recognition and suppression of stochastic pulsive interference (PI), due to, overlapping broad band frequency spectrum of PI and PD pulses. Therefore, on-line, on-site, PD measurement is hardly possible in conventional frequency based DSP techniques. We provide new methods to model and recognize the PD signal, on-line. The observed noisy PD signal is modeled as linear combination of systematic and random components employing probabilistic principal component analysis (PPCA). Being a natural signal, PD exhibits long-range dependencies. Therefore, we model the random part of the signal with fractional Brownian motion (fBm) process and pdf of the underlying stochastic process is obtained. The PD/PI pulses are assumed as the mean of the process and non-parametric analysis based on smooth FIR filter is undertaken. The method proposed by the Author found to be effective in recognizing and retrieving the PD pulses, automatically, without any user interference. [ABSTRACT FROM AUTHOR]

In this paper, a novel context-sensitive classification technique based on Support Vector Machines (CS-SVM) is proposed. This technique aims at exploiting the promising SVM method for classification of 2-D (or n-D) scenes by considering the spatial-context information of the pixel to be analyzed. The context-based architecture is defined by properly integrating SVMs with a Markov Random Field (MRF) approach. In the design of the resulting system, two main issues have been addressed: i) estimation of the observation term statistic (class-conditional densities) with a proper multiclass SVM architecture; ii) integration of the SVM approach in the framework of MRFs for modeling the prior model of images. Thanks to the effectiveness of the SVM machine learning strategy and to the capability of MRFs to properly model the spatial-contextual information of the scene, the resulting context-sensitive image classification procedure generates regularized classification maps characterized by a high accuracy. Experimental results obtained on Synthetic Aperture Radar (SAR) remote sensing images confirm the effectiveness of the proposed approach. [ABSTRACT FROM AUTHOR]

This paper proposes an unsupervised algorithm for learning a high-dimensional finite generalized Dirichlet mixture model. The generalized Dirichlet distribution offers high flexibility and ease of use. We propose a hybrid stochastic expectation maximization algorithm (HSEM) to estimate the parameters of the generalized Dirichlet mixture. The performance of our method is tested by applying it to the problems of image restoration. [ABSTRACT FROM AUTHOR]

In recent times, an increasing, worldwide effort has been devoted to the development of automatic personal identification systems that can be effective in a wide variety of security contexts. Palmprints have a number of unique advantages: they are rich in features such as principal lines, wrinkles, and textures and these provide stable and distinctive information sufficient for separating an individual from a large population. In this paper, we present a novel biometric authentication system to identify a person's identity by his/her palmprint. Being a robust and reliable system, it was tested by more than 8,000 palmprint images with very low false acceptance rate (0.02%), and a relative high genuine acceptance rate (98.83%). The whole authentication process is less than 1 second. Finally, some possible applications are discussed which could be benefited by using palmprint technology. [ABSTRACT FROM AUTHOR]

Symmetry is an important feature in vision. Several detectors or transforms have been proposed. In this paper we concentrate on a measure of symmetry. Given a transform S, the kernel SK of a pattern is defined as the maximal included symmetric sub-set of this pattern. It is easily proven that, in any direction, the optimal axis corresponds to the maximal correlation of a pattern with its flipped version. For the measure we compute a modified difference between respective surfaces of a pattern and its kernel. That founds an efficient algorithm to attention focusing on symmetric patterns. Keywords: Symmetry transforms, symmetry measure, correlation, feature extraction. [ABSTRACT FROM AUTHOR]

In this paper, we present a novel algorithm for 3D face recognition that is robust to the rotations and translations of the face models. Based on the Iterative Closest Point algorithm, a template based registration strategy is proposed for data normalization. Back-Propagation neural networks are then constructed to perform recognition tasks. The proposed algorithm is general purpose and can be applied for common 3D object recognitions. Experimental results illustrate that the algorithm is effective and robust. [ABSTRACT FROM AUTHOR]

This paper describes an application of CBR with decision tree induction in a manufacturing setting to analyze the cause for defects reoccurring in the domain. Abstraction of domain knowledge is made possible by integrating CBR with decision trees. The CID approach augments the recall and reuse done by CBR with statistical analysis that is focused towards the discovery of connections between specific defects and their possible causes. We show that this discovery also gives a pointer towards a corresponding corrective action. [ABSTRACT FROM AUTHOR]

The paper addresses the problem of constructing rough belief change functions. Contraction and revision postulates using the notion of rough consequence and rough consistency have been proposed in [2]. The base logic is a reasoning framework , that has a semantics of rough truth. We demonstrate here that functions satisfying the contraction postulates can be constructed through the method of partial meets [1]. As a result, a construction of rough revision functions is also obtained. [ABSTRACT FROM AUTHOR]

With the advent of efficient techniques for multi-objective evolutionary optimization (EMO), real-world search and optimization problems are being increasingly solved for multiple conflicting objectives. During the past decade, most emphasis has been spent on finding the complete Pareto-optimal set, although EMO researchers were always aware of the importance of procedures which would help choose one particular solution from the Pareto-optimal set for implementation. This is also one of the main issues on which the classical and EMO philosophies are divided on. In this paper, we address this long-standing issue and suggest an interactive EMO procedure which, for the first time, will involve a decision-maker in the evolutionary optimization process and help choose a single solution at the end. This study is the culmination of many year's of research on EMO and would hopefully encourage both practitioners and researchers to pay more attention in viewing the multi-objective optimization as an aggregate task of optimization and decision-making. [ABSTRACT FROM AUTHOR]

Measuring semantic nearness of documents is important for accurate information retrieval, automated text categorization and classification. Inspired by the observation that text documents contain semantically coherent set of ideas/topics, this paper presents the design and experimental evaluation of a method to represent a text document as a set of concepts. Based on this, we propose a method to measure semantic nearness of texts. Our method makes use of WordNet which is a lexico-semantic network of words. We bypass word sense disambiguation. In order to show the effectiveness of our representation of texts, we compare experimental results of text classification and clustering with the results of classification and clustering with standard techniques. [ABSTRACT FROM AUTHOR]

This paper deals with some new operators of genetic algorithms for solving the traveling salesman problem (TSP). These include a new operator called, "nearest fragment operator" based on the concept of nearest neighbor heuristic, and a modified version of order crossover operator. Superiority of these operators has been established on different benchmark data sets for symmetric TSP. Finally, the application of TSP with these operators to gene ordering from microarray data has been demonstrated. [ABSTRACT FROM AUTHOR]

Designing web sites is a complex problem. Adaptive sites are those which improve themselves by learning from user access patterns. In this paper we have considered a problem of index page synthesis for an adaptive website and framed it in a new type of Multi-Objective Optimization problem. We give a solution to index page synthesis which uses a popular clustering algorithm DBSCAN alongwith NSGA-II-an evolutionary algorithm-to find out best index pages for a website. Our experiments shows that very good candidate index pages can be generated automatically, and that our technique outperforms various existing methods such as PageGather, K-Means and Hierarchical Agglomerative Clustering. [ABSTRACT FROM AUTHOR]

In dealing with high-dimensional, large data, for the sake of abstract generation one resorts to either dimensionality reduction or cluster the patterns and deal with cluster representatives or both. The current paper examines whether there exists an equivalence in terms of generalization error. Four different approaches are followed and results of exercises are provided in driving home the issues involved. Keywords: Data Mining, prototype selection, frequent itemsets, clustering,feature selection. [ABSTRACT FROM AUTHOR]

In this paper, a new method for generating size-bounded clusters is proposed such that the cardinality of each cluster is less than or equal to a pre-specified value. First, set estimation techniques coupled with Rectangular Intersection Graphs are used to generate adaptive clusters. Then, the size-bounded clusters are obtained by using space partitioning techniques. The clusters can be indexed by a Kd-tree like structure for similarity queries. The proposed method is likely to find applications to Content Based Image Retrieval (CBIR). [ABSTRACT FROM AUTHOR]