Effects of Training Set Dimension on Recognition of Dysmorphic Faces with Statistical Classifiers

Şafak Saraydemir¹, Necmi Taşpınar², Osman Eroğul³, and Hülya Kayserili⁴

¹Department of Electronics Engineering, Turkish Military Academy, Turkey

²Department of Electrical and Electronics Engineering, Erciyes University, Turkey

³Biomedical Engineering Centre, Gülhane Military Medicine Academy, Turkey

⁴Department of Medical Genetics, İstanbul University Medicine Faculty, Turkey

 Abstract: In this paper, an evaluation using various training data sets for discrimination of dysmorphic facial features with distinctive information will be presented. We utilize Gabor Wavelet Transform (GWT) as feature extractor, K-Nearest Neighbor (KNN) and Support Vector Machines (SVM) as statistical classifiers. We analyzed the classification accuracy according to increasing dimension of training data set, selecting kernel function for SVM and distance metric for kNN. At the end of the overall classification task, GWT - SVM approach with Radial Basis Function (RBF) kernel type achieved the best classification accuracy rate as 97.5% with 400 images in training data set.

 Keywords: Dysmorphology, GWT, principal component analysis, face recognition, SVM, KNN.

Received February 26, 2013; accepted May 6, 2013

Full Text

Topical Web Crawling for Domain-Specific Resource Discovery Enhanced by Selectively Using Link-Context

Lu Liu1, 2, Tao Peng1, 2, 3, and Wanli Zuo1, 3

1College of Computer Science and Technology, Jilin University, China

2Department of Computer Science, University of Illinois at Urbana-Champaign, USA

3Key Laboratory of Symbol Computation and Knowledge Engineering of the Ministry of Education, China

  Abstract: To enable topical Web crawling, link-context is the critical contextual information of anchor text for retrieving domain-specific resources. While some link-contexts may misguide topical Web crawling and extract wrong Web pages, because several relevant anchor texts become irrelevant or several irrelevant anchor texts become relevant after calculating the relevance between the link-contexts and the feature terms of the specific topic. In view of above, this paper presents a heuristic-based approach by selectively using link-context and implements DOM tree to locate the anchor text. Unlike previous crawling algorithms, which only zero in on link-context and ignore whether it is really needed or not? Our method cares both link-context and evaluating its necessity to correctly use link-context to guide topical crawling. Accordingly, our topical crawler can retrieve more relevant Web pages. Experimental results indicate that this approach outperforms breadth-first, best-first, anchor text only, link-context both in harvest rate and target recall.

 Keywords: Topical crawling, domain-specific resource retrieving, selectively using link context, DOM tree

    Received November 17, 2012; accepted March 14, 2014 

Full Text

Arabic Text Classification Using K-Nearest Neighbour Algorithm

RoissAlhutaish and Nazlia Omar

Faculty of Information Science and Technology, Universiti Kebangsaan Malaysia, Malaysia

Abstract: Many algorithms have been implemented to the problem of Automatic Text Categorization (ATC). Most of the work in this area has been carried out on English texts, with only a few researchers addressing Arabic texts. We have investigated the use of the K-NN classifier, with anInew, Cosine,Jaccard, and Dice similarities, in order to enhance Arabic ATC. We represent the dataset as unstemmed and stemmed data; with the use of TREC-2002, in order to remove prefixes and suffixes. However, for statistical text representation, Bag-Of-Words (BOW) and character-level 3 (3-Gram) were used. In order to reduce the dimensionality of feature space, we used several feature selection methods. Experiments conducted with Arabic text showed that the K-NN classifier, with the new method similarity (Inew) 92.6% Macro-F1, had better performance than the K-NN classifier with Cosine, Jaccard, and Dice similarities. Chi-Square feature selection, with representation by Bag-Of-Words (BOW), led to the best performance over other feature selection methods using BOW and 3-Gram.

 Keywords: ATC, K-NN, similarity measures, feature selection methods.

Received May 3, 2012; accepted March 13, 2014

Full Text

Optimized Features Selection using Hybrid PSO-GA for Multi-view Gender Classification

Abstract: Gender classification is a fundamental face analysis task. In literature, the focus of most researchers has been on the face images acquired under controlled conditions. Real-world face images contain different illumination effects and variations in facial expressions and poses that make ge0nder classification more challenging task.  In this paper, we have proposed an efficient gender classification technique for real world face images (Labeled faces in the Wild). After extracting both global and local features using Discrete Cosine Transform (DCT) and Local Binary Pattern (LBP), we have fused these features. Proposed Algorithm provides support for variations in expressions and poses. To reduce the data dimensions, fused features are passed to hybrid PSO-GA that eliminates irrelevant features and results in optimized features. Support vector machine is trained and tested by using optimized features. Using this approach we have received a 98% accuracy rate. We are utilizing the minimum number of features so our technique is faster as compared to other state-of-the-art gender classification techniques.

  Keywords: Gender classification, feature extraction, pattern recognition, active shape model, real world face images.

Received February 6, 2013; accepted March 13, 2014

Full Text

Optimizing Ontology Alignments by Using NSGA-II

Xingsi Xue, Yuping Wang, and Weichen Hao

School of Computer Science and Technology, Xidian University, China

Abstract: In this paper, we propose a novel approach based on NSGA-II to address the problem of optimizing the aggregation of three different basic similarity measures (Syntactic Measure, Linguistic Measure and Taxonomy-based Measure), and get a single similarity metric. Comparing with conventional Genetic Algorithm, the proposed method is able to realize three goals simultaneously, i.e., maximizing the alignment recall, the alignment precision and the f-measure, and find the optimal solutions which could avoid bias to recall or precision value. Experiment results show that the proposed approach is effective.

 Keywords: ontology alignment, NSGA-II, aggregation of similarity measures.

Received June 15, 2012; accepted March 13, 2014

  Full Text

The Proposal of a Qualification Based Approach to Teach Software Engineering Course

Afnan Alsolamy and Rizwan Qureshi

Faculty of Computing and Information Technology, King Abdulaziz University, Saudi Arabia

Abstract: Software engineering students are facing many difficulties and challenging tasks when they approaching to industry. They suffer from inadequate experiences that are lacking in them to be skilled software engineers. This paper proposes a new approach to teach and train the students of software engineering course. The qualification approach (proposed in this paper) concentrates to establish a separate centre to provide a facility for students to understand real time environment similar to one in the industry. Furthermore, it focuses to live availability of real customers during software development. The approach also proposes to test each student using one of the personality tests. This will support to focus to develop a student’s skills in one area, which is related to their personality preference.

Keywords: Software engineering education, teaching, centre, personality test.

Received March 30, 2013; accepted September 11, 2013

Full Text

Chaos Genetic Algorithm Instead Genetic Algorithm

  Mohammad Javidi and Roghiyeh HosseinpourFard

Faculty of Mathematics and Computer, Shahid Bahonar University of Kerman, Iran

Abstract: Today the Genetic Algorithm (GA) is used to solve a large variety of complex nonlinear optimization problems. However, permute convergence which is one of the most important disadvantages in GA is known to increase the number of iterations for reaching a global optimum. This paper presents a new genetic algorithm based on chaotic systems to overcome this shortcoming,. We employ Logistic map and Tent map as two chaotic systems to generate chaotic values instead of the random values in Genetic Algorithm processes. The diversity of the Chaos Genetic Algorithm (CGA) avoids local convergence more often than the traditional GA. Moreover, numerical results show that the proposed method decreases the number of iterations in optimization problems and significantly improves the performance of the basic Genetic Algorithm. The idea of utilization of chaotic sequences for optimization algorithms is motivated by biological systems such as Particle Swarm Optimization (PSO), Ant Colony Algorithms (ACO) and bee colony algorithms and has the potential to improve ordinary genetic algorithms

Keywords: CGA, optimization problem, chaos evolutionary algorithm

Received November 12, 2012; accepted March 9, 2014

Full Text

A Biometric Based Secure Session Key Agreement Using Modified Elliptic Curve Cryptography

Usha Subramaniam1 and Kuppuswami Subbaraya2

1Department of Electrical & Electronic Engineering, Kongu Engineering College, India

2Professor & Principal, Kongu Engineering College, India

Abstract:  Protection of data and Network Security has been greatly researched. To enhance the security in the case of border control applications like E-Passport, conventional cryptographic concepts are integrated with biometrics. To avoid the intrusion of terrorists after the   terrorists attack of 9/11, many countries begin to issue E-Passport to their citizens contains biometric data like face, fingerprint and iris. The first generation E-Passport developed as per the standards and specifications of   International Civil Aviation Organization is confirmed to be lacking confidence and has numerous threats. The second generation E-Passport, was designed as per the mechanism of Extended Access Control also submits lots of threats especially in safety and confidentiality.  In this article, security enhanced mechanism based on   variation of Diffie-Hellman Key Agreement Protocol using Elliptic Curve Cryptography (ECC) between E-Passport and the Examination System is suggested. In the proposed method elliptic curve parameters A, B, and G are derived from the minutiae points of the fingerprint. From these parameters public key of E-Passport and session key between E-Passport and ES is generated. The security analysis of the proposed solution confirms the security goal of the biometric based system. The proposed protocol is developed using MATLAB (R2010b) tool.

Keywords: Active authentication, data originator and verifier, ECC, e-passport, examination system, passive authentication.

Received March 16, 2013; accepted January 31, 2014

 Full Text

Template Based Affix Stemmer for a Morphologically Rich Language

  Sajjad Khan1, Waqas Anwar1,2, Usama Bajwa1, and Wang Xuan2

1Department of Computer science, COMSATS Institute of Information Technology, Pakistan

2Harbin Institute of Technology Shenzhen Graduate School, China

Abstract Word stemming is one of the most significant factors that affect the performance of a Natural Language Processing (NLP) application such as information retrieval system, part of speech tagging, machine translation system and syntactic parsing. Urdu language raises several challenges to NLP largely due to its rich morphology. In Urdu language, stemming process is different as compared to that for other languages, as it not only depends on removing prefixes and suffixes but also on removing infixes. In this paper we introduce a template based stemmer that eliminates all kinds of affixes i.e. prefixes, infixes and suffixes, depending on the morphological pattern of the word. The presented results are excellent and this stemmer can prove to be very affective for a morphologically rich language.

 Keywords: Information retrieval, stemming, prefix, infix, suffix, exception lists.

Received October 19, 2012; accepted December 4, 2013

 Full Text

Combination of Feature Selection and Optimized Fuzzy Apriori Rules: The Case of Credit Scoring

Seyed Sadatrasoul, Mohammad Gholamian, and Kamran Shahanaghi

Faculty of Industrial Engineering, Iran University of Science and Technology, Iran

Abstract: Credit scoring is an important topic, and banks collect different data from their loan applicants to make appropriate and correct decisions. Rule bases are favourite in credit decision making because of their ability to explicitly distinguish between good and bad applicants. This paper uses four feature selection approaches as features pre-processing combined with fuzzy apriori. These methods are stepwise regression, CART, Correlation matrix and PCA. Particle Swarm is applied to find the best fuzzy apriori rules by searching different support and confidence. Considering Australian and German UCI and an Iranian bank datasets, different feature selections methods are compared in terms of accuracy, number of rules and number of features. The results are compared using T test; it reveals that fuzzy apriori combined with PCA creates a compact rule base and shows better results than the single fuzzy apriori model and other combined feature selection methods. Optimization outperformed FCFS and round-robin algorithms.

 Keywords: fuzzy apriori, feature selection, particle swarm, credit scoring.

Received September 7, 2012; accepted December 28, 2013

Full Text

Cloud Task Scheduling Based on Ant Colony Optimization

Medhat Tawfeek, Ashraf El-Sisi, Arabi Keshk, and Fawzy Torkey

Faculty of Computers and Information, Menoufia University, Egypt

Abstract: Cloud computing is the development of distributed computing, parallel computing and grid computing, or defined as the commercial implementation of these computer science concepts. One of the fundamental issues in this environment is related to task scheduling. Cloud task scheduling is an NP-hard optimization problem, and many meta-heuristic algorithms have been proposed to solve it. A good task scheduler should adapt its scheduling strategy to the changing environment and the types of tasks. In this paper a cloud task scheduling policy based on ant colony optimization algorithm compared with different scheduling algorithms FCFS and round-robin, has been presented. The main goal of these algorithms is minimizing the makespan of a given tasks set. Ant colony optimization is random optimization search approach that will be used for allocating the incoming jobs to the virtual machines. Algorithms have been simulated using Cloudsim toolkit package. Experimental results showed that cloud task scheduling based on ant colony optimization outperformed FCFS and round-robin algorithms.

 Keywords: Cloud computing, task scheduling, makespan, ant colony optimization, Cloudsim

Received July 3, 2013; accepted February 24, 2014

Full Text

A Multimodal Biometric System Based on Palmprint and Finger Knuckle Print Recognition Methods

Esther Perumal1 and Shanmugalakshmi Ramachandran2

1Associate Professor, Kathir College of Engineering, Coimbatore

2Associate Professor, Department of CSE, Government College of Technology, Coimbatore

Abstract: Biometric authentication is an effective method for automatically recognizing a person’s identity. In our previous paper, we have considered palm print for human authentication. Recently, it has been found that the Finger Knuckle Print (FKP), which refers to the inherent skin patterns of the outer surface around the phalangeal joint of one’s finger, has high capability to discriminate different individuals, making it an emerging biometric identifier. In this paper, the local convex direction map of the FKP image is extracted. Then the local features of the enhanced FKP are extracted using the Scale Invariant Feature Transform (SIFT), the Speeded Up Robust Features (SURF) and Frequency Feature. SIFT are formed by means of local patterns around key-points from scale space decomposed image. Feature vectors through SURF are formed by means of local patterns around key-points which are detected using scaled up filter. The Frequency range of pixel levels in each image is employed by using Empirical Mode Decomposition (EMD). For the authentication of FKP image, we used shortest distance between the query image and the database, to evaluate their similarity. Here we use PolyU FKP database images to examine the performance of the proposed system. The proposed biometric system is implemented in MATLAB and compared with the previous palm print human authentication system. For the same person, the matching score of the two methods are fused for the multimodal biometric recognition. The experimental results demonstrated the efficiency and effectiveness of this new biometric characteristic.

 Keywords: FKP, convex direction map, SIFT, SURF.

Received October 3, 2012; accepted August 2, 2013

Chaotic Image Encryption Using Modular Addition and Combinatorial Techniques 

 Sathishkumar Arthanari1, Mohamedmoideen Mastan1, and Boopathy BaganK2

1Dept of Electronics and Communication Engg, Sri Venkateswara College of Engineering, India

2Department of Electronics, Madras Institute of Technology, Anna University Chennai, India

Abstract: The image encryption is widely used to secure transmission of data in an open internet and internet works. For image based cryptosystems chaotic maps can be used as a key because of its nonlinear component. Due to sensitivity to initial conditions, chaotic maps have best alternative for designing dynamic permutation of the image based cryptosystem. A chaotic map is used to generate permutation matrix. An external secret key is used to derive the initial conditions for chaotic map. A pixel shuffling is used to expand diffusion property in the image and dissipate the high correlation among image pixels. The proposed algorithm is basically consisting of a pixel shuffling and modular addition plus permutation, which is a combination of block permutation, pixel permutation and value transformation. In general, diffusion and permutation is performed in a concurrent manner. These two methods are opened and operated alternatively in every round of encryption process at least four such chaotic sub keys are employed in every round of primitive encryption process. Decryption has the same structure, which operates in reverse order. Results of the various types of analysis are encouraging and imply that the proposed approach is able to adeptly trade off between the speed and protection. Hence it is suitable for the secure transmission of image, video and multi-media in real-time. The proposed algorithm is tested for different types of statistical analysis by examining their autocorrelation, cross correlation performance, measuring the histogram of the cipher image and the bit error probability for the received data in a communication system. 

Keywords: Image encryption, modular addition, chaotic maps, logistic map, socek, grp, cross and Block cipher.

Received September 5, 2012; accepted August 30, 2013