1 2251-6107 ICT Research Institute(ITRC) 89 Information Technology JNAM: a new detector proposed for marine radars Naseri Ali Jamshidi Jahan 1 9 2015 7 3 1 9 29 09 2018 29 09 2018 The proposed detector in this paper was obtained from a combination of adaptive and clutter-map detectors. Detection power of detectors has been studied in homogeneous and non-homogeneous environment (presence of interference targets) through MATLAB simulation software. The k-distribution proved to be the best option to display distribution of sea clutter, while k-distribution was assumed for sea clutter in simulation. On the other hand, CA-CFAR detector performed best in homogeneous conditions, and also the Ex-CFAR detector was suggested to improve the resistance of CA-CFAR detector against the interfering targets. The proposed detector performance was compared with these two detectors and with the ideal detector acquired from Marcum and Swerling equations, in homogeneous and non-homogeneous environment. Performance of the detectors in the presence of sharp clutter (which represents the fast clutter that make most difficult for CFAR detectors) and broad clutter (which represents the slow clutter) was investigated. Moreover, the performance of detectors in terms of data processing speed was discussed. Swerling ш model for oscillating signal target was received along with the detection of a signal pulse which was also considered in all simulation.
90 Information Technology 8 kbps Speech Coding using KLMS Prediction, Look-Ahead Adaptive Quantization and Pre-Emphasized Noise Reduction Alipoor Ghasem Savoji Mohammad Hassan 1 9 2015 7 3 11 19 29 09 2018 29 09 2018 A new scheme is developed, in this paper, within the framework of the ADPCM-based waveform coding technique for low bit rate encoding of speech signals. The essential feature of this scheme consists of replacing the commonly used linear filter with nonlinear processing based on kernel methods. Our previously reported study, conducted on various emerging kernel adaptive algorithms, shows the usefulness of the kernel LMS (KLMS) algorithm in this framework. However, two original strategies are incorporated into this scheme, in the current study, to further improve its performance. The first strategy is based on improving the adaptive scalar quantization of the residual samples by employing a look-ahead concept to find the best possible quantization levels using the Viterbi algorithm. The second strategy is to apply a pre-emphasized noise reduction filter. This filter is implemented in a closed-loop form along with an inverse filter, so as to minimize the destructive effects of the noise reduction filter. Simultaneous employment of these strategies in the main scheme with the nonlinear processing provided by the KLMS algorithm brings about a waveform encoder that reconstructs speech with PESQ measure of 2.5 at low bit rate of 1 bit per sample. 91 Information Technology Fast GLCM and Gabor Filters for Texture Classification of Very High Resolution Remote Sensing Images Mirzapour Fardin Ghassemian Hassan 1 9 2015 7 3 21 30 29 09 2018 29 09 2018 In the present research we have used gray level co-occurrence matrices (GLCM) and Gabor filters to extract texture features in order to classify satellite images. The main drawback of GLCM algorithm is its time-consuming nature. In this work, we proposed a fast GLCM algorithm to overcome the mentioned weakness of the traditional GLCM. The fast GLCM is capable of extracting approximately the same features as the traditional GLCM does, but in much less time (about 200 times faster). The other weakness of the traditional GLCM is its lower accuracy in the regions near the class borders. Since features extracted using Gabor filters are more accurate in boundary regions, we combined Gabor features with GLCM features. In this way we could compensate the latter mentioned weakness of GLCM. Experimental results show good capabilities of the proposed fast GLCM and the feature fusion method in classification of very high resolution remote sensing images. 92 Information Technology Using Hybrid of Eigenface and Imperialist Competition Algorithm for Feature Selection in Face Recognition Yousefi Darestani Mohammad Reza Sheikhan Mansour Khademi Maryam 1 9 2015 7 3 31 42 29 09 2018 29 09 2018 Design of a robust Face Recognition (FR) system is greatly affected due to varying illumination and pose conditions. The accuracy of FR system can be increased by normalizing and compensating the illumination variations in the pre-processing stage. To improve the robustness of FR systems against illumination variations, a method is proposed in this study which is based on Contourlet Transform (CT), hybrid of Principal Component Analysis (PCA) and Imperialist Competition Algorithm (ICA) techniques for feature reduction, and an ICA-optimized Multi-Layer Perceptron (MLP) classifier. First, each face is decomposed using the CT. So, the contourlet coefficients of low and high frequency in different scales and various angles are obtained. The frequency coefficients are employed as a feature vector for further processing. The PCA is then used to reduce the dimensionality of the feature vector. The ICA is also exploited to search the feature space for an optimal feature subset. Then, the reduced-size feature vector is applied to the face classifier that is based on MLP neural network whose structure and learning rate are optimized by ICA. The proposed method is robust to variation of imaging conditions and pose variations. The proposed technique provides better results when tested on ORL and Extended Yale-B databases as compared with other existing techniques such as hybrid model based on discrete wavelet transform and PCA (in terms of precision, sensitivity, and accuracy) and different state-of-the-art methods. 95 Information Technology Discovering Influencers for Spreading in Weighted Networks Rezaei Zahra Tarokh Mohammad Jafar 1 9 2015 7 3 43 51 29 09 2018 29 09 2018 Identifying the influential nodes in networks is an important issue for efficient information diffusion, controlling rumors and diseases and optimal use of network structure. The degree centrality which considers local topology features, does not produce very reliable results. Despite better results of global centrality such as betweenness centrality and closeness centrality, they have high computational complexity. So, we propose semi-local centrality measure to identify influential nodes in weighted networks by considering node degree, edges weight and neighboring nodes. This method runs in O(n(k)2). So, it is feasible for large scale network. The results of applying the proposed method on weighted networks and comparing it with susceptible-infected-recovered model, show that it performs good and the influential nodes are generated by our method can spread information well. 93 Information Technology Resource Reservation in Grid Networks based on Irregular Cellular Learning Automata Rezaei Sara Khademzadeh Ahmad Sheikhan Mansour 1 9 2015 7 3 53 61 29 09 2018 29 09 2018 Computing infrastructures that are based on grid networks have been recognized as a basis for new infrastructures of distributed computing. Providing appropriate mechanisms for scheduling and allocating resources to user’s requests in these networks is considered very important. One of the current issues in the grid networks is how to ensure the precise timing of executing requests sent by users, especially requests that have deadlines and also co-allocation requests. The resource reservation has been mainly developed to address this problem in the grid systems. On the other hand, models based on the cellular automata have advantages such as lower processing complexity, configurability of the cells, and the ability of predicting future conditions. In this study, an efficient model based on irregular cellular learning automata (ICLA) is presented for the task of resource reservation. The proposed model was simulated on a network with random topology structure. The performance of proposed method was compared with two well-known algorithms in this field. The experimental results showed increased efficiency in the resource utilization, decreased process execution delays, and reduced rate of request rejection. 94 Information Technology Density-Based K-Nearest Neighbor Active Learning for Improving Farsi-English Statistical Machine Translation System Bakhshaei Somayeh Safabakhsh Reza Khadivi Shahram 1 9 2015 7 3 63 72 29 09 2018 29 09 2018 Labeled data are useful resources for different application in different fields like image processing, natural language processing etc. Producing labeled data is a costly process. One efficient solution for alleviating the costly process of annotating data is managing the sampling process. It is better to query for essential samples instead of a group of unnecessary ones. Active learning (AL) attempts to overcome the labeling bottleneck by sending queries for unlabeled instances to be labeled with the help of an annotator. This technique is applied to Natural Language Processing (NLP) especially in Statistical Machine Translation (SMT) tasks that we also focus on in this work. In Statistical Machine Translation, parallel corpora are scarce resources, and AL is a way of solving this problem. It attempts to alleviate the costly process of data annotating by sending queries just for translation of the most informative sentences which are essential for system improvement. The contribution of our work is proposing a new approach in AL for selecting sentences through a soft decision making process. In this algorithm, in addition to scoring sentences according to their information, the distribution of the space of unlabeled data is also considered. Each sentence (either labeled or unlabeled) changes to a vector of feature scores. Then each new coming sentence is observed in the feature space and gets two probabilities: how probable it is to be either labeled or unlabeled. These probabilities are calculated according to the position of new instance related to its labeled and unlabeled neighbors. We have applied the proposed model for improving training corpus of a SMT system. Also Farsi-English language pairs are selected as the base-line SMT system. We have sampled the best sentences that can improve the quality of our SMT system and send query for their translations. In this way the costly approach of making parallel corpus is alleviated. Finally, our experiments show significant improvements for sampling sentences by soft decision making in comparison to the random sentence selection strategy.