Journal of Advances in Computer Research
Volume:5 Issue: 4, Autumn 2014

Zinc is the first modelling language which supports solver and technique independence. This means that a high level conceptual model can be automatically mapped into an appropriate low level design model for a specific solver or technique. To date, Zinc uses three different techniques to solve a model: Constraint Programming (CP), Local Search (LS), and Mixed Integer Programming (MIP). In this way, modellers can examine all solving techniques for their models and see which one gives them the best result. MIP solvers can only accept linear models. Therefore, to map a conceptual model to MIP solvers, the model must be linearized first. In this paper we explain the techniques used in Zinc to linearize high level data structures and expressions which may be appeared in a conceptual model. As a result, modellers can benefit of expressive modelling using nonlinear expressions as well as efficiency of MIP solvers.

Removal of artifacts from bio-signals is a necessary step before automatic processing and obtaining clinical information. Recently, many applications of Empirical Mode Decomposition (EMD) on biomedical researches have been presented that artifact reduction from bio-signals is one of them. EMD separates a time series into finite numbers of its individual oscillations, which are called intrinsic mode function (IMF). The process of the IMF extraction from a signal is known as the sifting process. The main issue during sifting process is to select an appropriate criterion for stopping IMF extraction when the process reaches the artifact components. In this paper, we try to investigate mean power frequency (MPF) for stopping sifting process, in case of low frequency artifact reduction from ECG and EMG signals. In order to evaluate effectiveness of the proposed index during sifting process, reduction of baseline noise from electrocardiogram signals (ECG) and ECG artifact from electromyogram signals (EMG) have been investigated. The obtained results indicate MPF can be considered as an acceptable criterion to stop sifting process during low frequency artifact elimination.

The fusion of medical images is very useful for clinical application. Generally, the PET image indicates the function of tissue and the MRI image shows the anatomy of tissue. In this article we fused MRI and PET images and the purpose is adding structural information from MRI to PET image. The images decomposed with Curvelet Transform, and then two images fused with applying fusion rules. We used MATLAB software for fused images and evaluated the result. The data set consists 34 images of color PET images and high resolution MRI images. The brain images are classified into two groups, normal (Coronal, Sagittal and Transaxial) and Alzeimer’s disease dataset images. Finally we used visual and quantitative criteria to evaluate the fusion result. In quantitative evaluation we used entropy, discrepancy and overall performance. Results show the amount of entropy, achieved by the proposed method, was the highest and amount of discrepancy and overall performance was the lowest. The small amount of discrepancy, overall performance and high amount of entropy means high quality.

Finger vein is one of the most fitting biometric for identifying individuals. In this paper a new method for finger vein recognition is proposed. First the veins are extracted from finger vein images by using entropy based thresholding. In finger vein images the veins are appeared as dark lines. The method extracts veins as well, but the images are noisy, that means in addition to the veins they have some short and long lines. Then radon transformation are applied to segmented images. The Radon transform is not sensitive to the noise in the images due to its integral nature, so in comparison with other methods is more resistant to noise. For extracting dominant features from finger vein images, common spatial patterns (CSP) is applied to the blocks of radon transformation. Finally the data classified by using nearest neighbor (1-NN) and multilayer perceptron (MLP) neural network. The research was performed on the Peking University finger vein dataset. Experimental results show that 1-NN using CSP, with detecting rate 99.6753%, against MLP is most appropriate for finger vein recognition.

In recent years, the needs of the Internet are felt in lives of all people. Accordingly, many studies have been done on security in virtual environment. Old technics such as firewalls, authentication and encryption could not provide Internet security completely; So, Intrusion detection system is created as a new solution and a defense wall in cyber environment. Many studies were performed on different algorithms but the results show that using machine learning technics and swarm intelligence are very effective to reduce processing time and increase accuracy as well. In this paper, hybrid SVM and ABC algorithms has been suggested to select features to enhance network intrusion detection and increase the accuracy of results. In this research, data analysis was undertaken using KDDcup99. Such that best features are selected by Support vector machine, then selected features are replaced in the appropriate category based on artificial bee colony algorithm to reduce the search time, increase the amount of learning and improve the authenticity of intrusion detection. The results show that the proposed algorithm can detect intruders accurately on network up to 99.71%.

In this paper, a Robust Integral of the Sign Error (RISE) feedback controller is designed for a Rigid-Link Electrically Driven (RLED) robot manipulator actuated by direct current DC motor in presence of parametric uncertainties and additive disturbances. RISE feedback with implicitly learning capability is a continuous control method based on the Lyapunov stability analysis to compensate an additive bounded disturbance and linear in the parametric (LP) and non-linear in parametric (non-LP) uncertain dynamics through the use of a sufficiently large gain multiplied by an integral signum term. A proper selection of controller gains in predefined permitted areas for gains leads to reduce convergence time, control effort and improve the performance. The Bees Algorithm that is a search procedure inspired by the foraging behaviour of honey bees is used to tune the parameters of the controller to achieve the convergence. The performance of the proposed controller is compared with a PD and neural network based controllers. Simulation results of a two Rigid-Link Electrically-Driven Robot Manipulator show the advantages of the proposed controller in terms of the transient and steady state performances in comparison with some conventional controllers.

Speaker verification is the process of accepting or rejecting claimed identity in terms of its sound features. A speaker verification system can be used for numerous security systems, including bank account accessing, getting to security points, criminology and etc. When a speaker verification system wants to check the identity of individuals remotely, it confronts problems such as noise effect on speech signal and also identity falsification with speech synthesis. In this system, we have proposed a new speaker verification system based on Multi Model GMM, called SV-MMGMM, in which all speakers are divided into seven different age groups, and then an isolated GMM model for each group is created; instead of one model for all speakers. In order to evaluate, the proposed method has been compared with several speaker verification systems based on Naïve, SVM, Random Forest, Ensemble and basic GMM. Experimental results show that the proposed method has so better efficiency than others.

Service Availability is important for any organization. This has become more important with the increase of DoS attacks. It is therefore essential to assess the threat on service availability. We have proposed a new model for threat assessment on service availability with a data fusion approach. We have selected three more important criteria for evaluating the threat on service availability and used anomaly detection algorithms to evaluate the network behavior. Anomaly of each parameter over time was measured based on its past behavior. The results of each algorithm were aggregated using the order weighted average (OWA) and finally using fuzzy inference system (FIS), threat has been calculated. We have evaluated our proposed model with data from a web server monitoring. The results show that it can provide network administrator with useful information about the status of service availability and help them to reduce threats and losses due to their actual activation.