NASHRIYYAH -I MUHANDISI -I BARQ VA MUHANDISI -I KAMPYUTAR -I IRAN, B- MUHANDISI -I KAMPYUTAR
Year:2016 | Volume:14 | Issue:1|Papers Count:8

This paper addresses the cooperative object transportation by a multi robot distributed system, which is a difficult problem due to path planning and robot cooperation challenges. In this problem, a number of robots should transport an object to a goal point safely while avoiding obstacles and utilizing a proper coordination and cooperation mechanism. The proposed method has a two-layer structure which benefits from both centralized and decentralized architectures. The global level takes advantage of full knowledge of environment to plan an optimal path using the new Optimally-Connect Random Tree (ORT) method, and the local level performs some local processes to reduce the system’s overall processing load and cost and increase its robustness. The required coordination between the robots is realized via radio communication, and for local path planning of the robots a combination of potential fields and TangentBug algorithms has been used. The proposed method has been implemented on multiple KUKA youBot mobile manipulators in the Webots simulation software, and its performance has been evaluated through various experimentations and the results of implementing and comparing the ORT and Rapidly-exploring Random Trees (RRT) showed the advantage of the proposed method.

More than two decades ago, the balanced scorecard method was proposed to control and monitor the strategy of organizations. The most important outcome of this method is the strategy map. The causal relations among strategic goals (SGs) are established in this map which can help managers in decision making process. To have a precise strategy map, it is necessary to estimate the strengths of each causal relation correctly. So, the estimation of causal coefficients has attracted research interest in forming strategy maps. In this way, DEMATEL and Delphi are two well-known methods that are based on the experts’ opinion. However, these opinions are not exact in the complex business fields; so, the computational intelligence (CI) algorithms have been employed for more precise estimation of causality coefficients. In this study, the relations among SGs and their coefficients have been provided by the experts of a banking institution as the input of the proposed method. The main purpose of this study is to improve the precision of causal coefficients using a CI-based algorithm. For this purpose, the strategy map is decomposed into multiple fuzzy cognitive maps (FCMs) and then, the gravitational search algorithm (GSA) is employed for FCM training. In this way, two objective functions are used for determining the optimal value of causality coefficients. The first objective function is employed for reducing error in the prediction of SG realization percentages. The second objective function keeps causal coefficients in the intervals determined by the experts. Experimental results show that the total error of proposed model is lower than the expert-based model. In addition, GSA performs better than the following algorithms in finding the global optimum point in this real-world case study: particle swarm optimization and ant colony optimization.

The structure of a Bayesian network represents a set of conditional independence relations that hold in the domain. Learning the structure of the Bayesian network model that represents a domain can reveal in sights into its underlying causal structure. Automatically learning the graph structure of a Bayesian network is a challenge pursued within artificial intelligence studies. In this paper, a new algorithm based on learning automata is proposed for learning the structure of the Bayesian networks. In this algorithm, automata is used as a tool for searching in structure’s space (DAG’s space) of the Bayesian networks. The mathematical behavior of the proposed algorithm is studied.

This paper presents a new method for multi human face detection from frontal view in internet images with complex background. The main goal is to reduce false acceptance error rate using feed forward back propagation multilayer perceptron neural network and Gabor energy feature in the frequency domain. In the proposed method, the false acceptance error extremely decreased using a combination of three operations; introducing a new preprocessing algorithm to increase the quality of Gabor energy feature, performing two step monitoring on the input and output images, and utilizing three indexes of facial components recognition in Gabor energy output. In this paper, a new image database namely RFD is collected from internet images including 583 non repetitive face images and 9961 non face images with size of 192×168. The face detection accuracy of the proposed method on RFD images is 88.16% with false acceptance rate of 0.48% or 48 false acceptances only, while Viola-Jones algorithm has 124 false acceptances. Therefore, the false acceptance error of the proposed method has reduced by 2.5 times compared to that of Viola-Jones algorithm.

In this study, an ensemble classification method, based on negative correlation learning, is used for holistic recognition of handwritten words with limited vocabulary. In this method, training data set, after preprocessing and feature extraction, is applied to the base Multilayer Perceptron classifiers. These classifiers are trained by negative correlation learning to make them diverse. Features extracted from a test input are applied to the base classifiers, which produce somehow diverse outputs. By combining these outputs, the final output of the system is obtained. For experiments, three feature sets based on zoning, gradient image and contour chain code are extracted from the images. In experiments, performed on 775 images of 31 Province centers from "Iranshahr" dataset, when gradient-based features were used to train 6 Multilayer Perceptron classifiers by negative correlation, by Fusion the outputs of these classifiers through voting, an average recognition rate of 96.10 percent is achieved.

Today, botnets are known as one of the most important threats against Internet infrastructure. A botnet is a network of compromised hosts (bots) remotely controlled by a so-called botmaster through one or more command and control (C&C) servers. Since DNS is one of the most important services on Internet, botmasters use it to resistance their botnet. By use of DNS service, botmasters implement two techniques: IP-flux and domain-flux. These techniques help an attacker to dynamically change C&C server addresses and prevent it from becoming blacklisted. In this paper, we propose a reputation system used a clustering method and DNS traffic for online fluxing botnets detection .we first cluster DNS queries with similar characteristics at the end of each time period. We then identify hosts that generate suspicious domain names and add them to a so-called suspicious group activity matrix. We finally calculate the negative reputation score of each host in the matrix and detect hosts with high negative reputation scores as bot-infected. The experimental results show that it can successfully detect fluxing botnets with a high detection rate and a low false alarm rate.

The hyperspectral images allow us to discriminate between different classes with more details. There are lots of spectral bands in hyperspectral images. On the other hand, the limited number of available training samples causes difficulties in classification of high dimensional data. Since the gathering of training samples is hard and time consuming, feature reduction can considerably improve the performance of classification. So, feature extraction is one of the most important preprocessing steps in analysis and classification of hyperspectral images. Feature extraction methods such as LDA have not good efficiency in small sample size situation. A supervised feature extraction method is proposed in this paper. The proposed method, which is called cluster space linear discriminant analysis (CSLDA), without obtaining the label of testing samples and just with doing a clustering on testing data, finds the relationship between training and testing samples. Then, it uses the power of unlabeled samples together with training samples for estimation of within-class and between-class scatter matrices. The CSLDA improves the classification accuracy particularly in multimodal hyperspectral data. The experimental results on urban and agriculture hyperspectral images show the better performance of CSLDA compared to popular feature extraction methods such as LDA, GDA, and NWFE using limited number of training samples.

This paper develops a new method of onset detection for the Tar, a traditional Iranian musical instrument. The proposed method is based on both types of pitch and energy features and an adaptive peak picking algorithm is utilized for primary onset detection. An improved template matching method is used to detect fundamental frequencies and finally, onsets are tagged based on primary onsets and fundamental frequencies. This step is especially useful to detect the reaz, repeatedly played notes with the same frequency and short durations. For the evaluation of the method, a data set with predetermined onsets was produced and the results were compared with an energy based method explained in terms of F measure.