Abstract Facial analysis is one of the challenges of a Human Machine Interactive (HMI) system. It includes face alignment, pose estimation, face recognition, facial expression detection etc. This paper proposes an efficient optimization technique by the hybridization of genetic algorithm (GA) with gradient descent (GD) to make a robust, efficient and real time face alignment system. It employs 2.5D Active Appearance... Facial analysis is one of the challenges of a Human Machine Interactive (HMI) system. It includes face alignment, pose estimation, face recognition, facial expression detection etc. This paper proposes an efficient optimization technique by the hybridization of genetic algorithm (GA) with gradient descent (GD) to make a robust, efficient and real time face alignment system. It employs 2.5D Active Appearance Model (AAM) for the face search, in which the face model is generated by taking 3D landmarks and 2D texture of the facial image. Facial large lateral movements requires to optimize 6DOF (Degrees of Freedom) pose parameters which make the facial search space of AAM nonconvex. This nonconvex multidimensional search space requires an efficient optimization methodology. Gradient descent also known as deterministic optimization method can optimize the parameters as long as the search space remains convex. However in non-convex search space GA is able search face globally while GD can help GA to search face locally. In other words exploitation property of GD and exploration property of GA are combined to make an efficient optimization method. For this hybrid optimization we propose a gradient operator in GA, which functions in conjunction with the existing genetic operator of mutation. Thus it does not increase the computational cost of the system. We compare it with classical search algorithm by gradient descent and the hybrid optimization of GA with Simplex. Facial databases of SUPELEC’08, Pointing’04 and synthetic characters comprising of different facial poses are analyzed. Simulation results validate the efficiency, accuracy and robustness achieved. Read More Read Less

Abstract We identify and explore an Information Retrieval paradigm called Query-By-Multiple-Examples (QBME) where the information need is described not by a set of terms but by a set of documents. Intuitive ideas for QBME include using the centroid of these documents or the well-known Rocchio algorithm to construct the query vector. We consider this problem from the perspective of text classification,... We identify and explore an Information Retrieval paradigm called Query-By-Multiple-Examples (QBME) where the information need is described not by a set of terms but by a set of documents. Intuitive ideas for QBME include using the centroid of these documents or the well-known Rocchio algorithm to construct the query vector. We consider this problem from the perspective of text classification, and find that a better query vector can be obtained through learning with Support Vector Machines (SVMs). For online queries, we show how SVMs can be learned from one-class examples in linear time. For offline queries, we show how SVMs can be learned from positive and unlabeled examples together in linear or polynomial time, optimising some meaningful multivariate performance measures. The effectiveness and efficiency of the proposed approaches have been confirmed by our experiments on four real-world datasets. Read More Read Less

Abstract Permanent growth of scientific publications makes bibliographic databases and digital libraries widespread. At the same time they are an object of research in their own right. In this paper we address the question of ”where do the authors come from?” via language identification of the author names. This is a two-steps process which involves primary classification based on the statistical... Permanent growth of scientific publications makes bibliographic databases and digital libraries widespread. At the same time they are an object of research in their own right. In this paper we address the question of ”where do the authors come from?” via language identification of the author names. This is a two-steps process which involves primary classification based on the statistical models of languages, and classification refinement achieved with the analysis of the co-author network built from the bibliographic records. A system for automatic language identification presented here handles 14 different languages and requires no dictionary of names for traing. The statistical models are built from the general purpose corpora for all Western European, Chinese, Japanese and Turkish languages. The system is fine tuned to achieve precision and recall above 90% for many languages, and provides better performance than some other systems aiming at the language identification of personal names. Tests on the DBLP data set have shown that the extension of the language model with the co-author network helps to improve classification results, especially in cases of closely related languages and mixed names. They have also demonstrated the usability of the system in applications such as data cleaning and trends detection. Read More Read Less

Abstract The growing advancement of wireless communication technologies, mobiles devices, and sensors enable new oportunities for users to share resources anytime and anywhere, making access control a critical concern. Such sensor-rich environments should be able to offer distributed and dynamic access to resources according to context information, such as user’s location and time. In this scenario, the access control decisions should... The growing advancement of wireless communication technologies, mobiles devices, and sensors enable new oportunities for users to share resources anytime and anywhere, making access control a critical concern. Such sensor-rich environments should be able to offer distributed and dynamic access to resources according to context information, such as user’s location and time. In this scenario, the access control decisions should be based also on context information. However, as context information represents real-world situations, it is associated with certain quality indicators, such as up-to-dateness and precision, which can directly affect the access control decisions. Furthermore, enforcing user’s privacy policies on context information can decrease the quality of context (QoC) information used for making access control decisions. Quality of context has in fact a strong impact on the suitable behavior of contextbased access control models. Using context information with inadequate quality may increase the probability of incorrect access control decisions. Therefore, it is important to take into account also the QoC while enforcing context-based access control policies. This paper proposes a Context-Based Access Control Model that takes into account context information and its QoC indicators to grant and to adapt access permissions on resources for Ubiquitous Computing Environments. A new module to evaluate QoC indicators was integrated in our context management architecture that is being used by proposed access control model. Moreover, we present an application called PhotoMap that uses the proposed model for controlling access to photos. Read More Read Less

Abstract Multidimensional arrays storing multidimensional data in MOLAP are usually very sparse. They also suffer from the problem that the time consumed in sequential access to array elements heavily depends on the dimension along which elements are accessed. This dimension dependency would be caused by arranging multidimensional data in linear order on secondary storage and the logical proximity among the multidimensional... Multidimensional arrays storing multidimensional data in MOLAP are usually very sparse. They also suffer from the problem that the time consumed in sequential access to array elements heavily depends on the dimension along which elements are accessed. This dimension dependency would be caused by arranging multidimensional data in linear order on secondary storage and the logical proximity among the multidimensional data would be lost in physical storage. This situation can be alleviated by dividing the whole array into the set of subarrays called “chunks”. However, further dimension dependency would be caused, unless the compressed chunks are arranged judiciously in the page buffer. The difference among the dimension cardinalities could also cause dimension dependency; slice operation along a dimension of large cardinality tends to consume much time. We will alleviate these two kinds of dimension dependency by introducing the notion of an “extended chunk”. Extended chunks can contribute to preserve the logical proximity by adapting flexibly to the usual situation where data densities are very low and are not uniformly distributed. Based on extended chunks, we will propose secondary storage schemes for multidimensional datasets using a space-filling curve such as Z-curve. The evaluation result shows that the proposed storage schemes exhibit better performance while alleviating the dimension dependency. Read More Read Less

Abstract The aim of this paper is to provide a better understanding on the role of interoperability in eGovernment applications. The focus of interest is on eGovernment applications involving more than one agency or heterogeneous technical infrastructures. The main research objective is to identify a set of critical factors that influence eGovernment interoperability. To fulfil this objective we reviewed the relevant... The aim of this paper is to provide a better understanding on the role of interoperability in eGovernment applications. The focus of interest is on eGovernment applications involving more than one agency or heterogeneous technical infrastructures. The main research objective is to identify a set of critical factors that influence eGovernment interoperability. To fulfil this objective we reviewed the relevant literature and examined 18 case studies of European eGovernment applications where interoperability has a significant role. The results enabled us to identify a number of critical factors categorized as technical, semantic, organizational, political, legal, managerial and economic. We believe that the results of this research can be used by practitioners and researchers alike. Practitioners may use the results as guidance when developing their own applications. Researchers can further extend these results, for example, by attempting to better understand current trends, by identifying additional critical factors and by investigating the relationships between factors. Read More Read Less

Abstract Conceptual Universal Database Language (CUDL) is a new language designed to manage dynamic database environments, which conform to the Frame DataBase model (FDB). FDB is a generic database model (oversubscribe both the Entity-Attribute-Value and the Nested Relational). CUDL is not only an FDB database language but it is mainly an agent that provides an abstraction level superior to the logical... Conceptual Universal Database Language (CUDL) is a new language designed to manage dynamic database environments, which conform to the Frame DataBase model (FDB). FDB is a generic database model (oversubscribe both the Entity-Attribute-Value and the Nested Relational). CUDL is not only an FDB database language but it is mainly an agent that provides an abstraction level superior to the logical level. CUDL permits the users to conceive a database schema where single fields (tags) accept repetitions (list of values), entertain subfields and also permit to entertain an entire table in the place of a single field. In this paper we investigate the conditions, used in both retrieval and update CUDL statements, and focalize especially on the range of affection of the update statements. This becomes necessary since the CUDL abstraction level reveals the differentiation between frame objects, tag repetitions and subfield repetitions and consequently the update statements should clearly define where the data modifications occur. This need becomes heavier where some frame object verifies two (or more) disjunctive conditions and these conditions specify different range of affection inside the frame. We provide two solutions where the second one is an improvement of the first one but it demands more refined statements and more instructed / skilled users. Some constructive discussion is also provided. Read More Read Less