Work place: Computer Engineering Department, Islamic University, P. O. Box 108, Gaza, Palestine
Research Interests: Image Processing, Pattern Recognition, Computer Vision, Artificial Intelligence
Ibrahim S. I. Abuhaiba is a professor at the Islamic University of Gaza, Computer Engineering Department.He obtained his Master of Philosophy and Doctorate of Philosophy from Britain in the field of document understanding and pattern recognition. His research interests include artificial intelligence, computer vision, image processing, document analysis and understanding, pattern r recognition, information security, and computer networks. Prof. Abuhaiba published tens of original contributions in these fields in well-reputed international journals and conferences.
DOI: https://doi.org/10.5815/ijisa.2017.04.05, Pub. Date: 8 Apr. 2017
The objective of this research is to improve Arabic text documents classification by combining different classification algorithms. To achieve this objective we build four models using different combination methods.
The first combined model is built using fixed combination rules, where five rules are used; and for each rule we used different number of classifiers. The best classification accuracy, 95.3%, is achieved using majority voting rule with seven classifiers, and the time required to build the model is 836 seconds.
The second combination approach is stacking, which consists of two stages of classification. The first stage is performed by base classifiers, and the second by a meta classifier. In our experiments, we used different numbers of base classifiers and two different meta classifiers: Naïve Bayes and linear regression. Stacking achieved a very high classification accuracy, 99.2% and 99.4%, using Naïve Bayes and linear regression as meta classifiers, respectively. Stacking needed a long time to build the models, which is 1963 seconds using naïve Bayes and 3718 seconds using linear regression, since it consists of two stages of learning.
The third model uses AdaBoost to boost a C4.5 classifier with different number of iterations. Boosting improves the classification accuracy of the C4.5 classifier; 95.3%, using 5 iterations, and needs 1175 seconds to build the model, while the accuracy is 99.5% using 10 iterations and requires 1966 seconds to build the model.
The fourth model uses bagging with decision tree. The accuracy is 93.7% achieved in 296 seconds when using 5 iterations, and 99.4% when using 10 iteration requiring 471 seconds. We used three datasets to test the combined models: BBC Arabic, CNN Arabic, and OSAC datasets. The experiments are performed using Weka and RapidMiner data mining tools. We used a platform of Intel Core i3 of 2.2 GHz CPU with 4GB RAM.
The results of all models showed that combining classifiers can effectively improve the accuracy of Arabic text documents classification.
DOI: https://doi.org/10.5815/ijisa.2016.06.04, Pub. Date: 8 Jun. 2016
Author attribution is the problem of assigning an author to an unknown text. We propose a new approach to solve such a problem using an extended version of the probabilistic context free grammar language model, supplied by more informative lexical and syntactic features. In addition to the probabilities of the production rules in the generated model, we add probabilities to terminals, non-terminals, and punctuation marks. Also, the new model is augmented with a scoring function which assigns a score for each production rule. Since the new model contains different features, optimum weights, found using a genetic algorithm, are added to the model to govern how each feature participates in the classification. The advantage of using many features is to successfully capture the different writing styles of authors. Also, using a scoring function identifies the most discriminative rules. Using optimum weights supports capturing different authors’ styles, which increases the classifier’s performance. The new model is tested over nine authors, 20 Arabic documents per author, where the training and testing are done using the leave-one-out method. The initial error rate of the system is 20.6%. Using the optimum weights for features reduces the error rate to 12.8%.[...] Read more.
DOI: https://doi.org/10.5815/ijwmt.2015.02.03, Pub. Date: 18 Apr. 2015
In wireless sensor networks, the issue of nodes localization has taken a wide area of research. Most applications need to know the position of sensor nodes for reasons of optimal and fast data routing. In this paper, a new distributed localization algorithm based on Self Organizing Maps (SOMs) is proposed to determine the location of a node in a wireless sensor network.
The proposed algorithm is classified as a range-free algorithm which uses only the connectivity information between nodes without the need to measure the time of arrival or signal strength as range-based algorithms require. It utilizes the neighborhood information and the well-known anchors' positions to calculate the estimated locations of nodes. Our algorithm is made up of two main stages. The initial estimated locations of nodes are calculated in the initialization stage, and fed to the learning stage in which a SOM is used to calculate the final estimated locations of nodes.
By using the neighborhood information at the first stage, the algorithm has significantly reduced the SOM learning time and the number of iterations to converge. On the other hand, starting with real data rather than random data maximized the accuracy of the resulted locations. Furthermore, the distributed implementation of the algorithm highly alleviated the pressure on the wireless nodes which are characterized with low power and limited capabilities.
The proposed algorithm has been implemented using MATLAB software and experimented by deploying different number of nodes in a specific area with different communication radio ranges. Extensive simulations evidently verified the performance of the algorithm and achieved a very good accuracy. Moreover, the algorithm proved its effectiveness with a lower average error and lower number of iterations compared to other related algorithms.
DOI: https://doi.org/10.5815/ijitcs.2014.06.10, Pub. Date: 8 May 2014
In this research, new ideas are proposed to enhance content-based image retrieval applications by representing colored images in terms of its colors and angles as a histogram describing the number of pixels with particular color located in specific angle, then similarity is measured between the two represented histograms. The color quantization technique is a crucial stage in the CBIR system process, we made comparisons between the uniform and the non-uniform color quantization techniques, and then according to our results we used the non-uniform technique which showed higher efficiency.
In our tests we used the Corel-1000 images database in addition to a Matlab code, we compared our results with other approaches like Fuzzy Club, IRM, Geometric Histogram, Signature Based CBIR and Modified ERBIR, and our proposed technique showed high retrieving precision ratios compared to the other techniques.
DOI: https://doi.org/10.5815/ijigsp.2013.05.01, Pub. Date: 28 Apr. 2013
In many applications, images are sensitive to an extent such that any modification in it could lead to serious problems. For example, hiding any portion of a medical image could lead to a misdiagnosis. Thus, detecting forgery in images is a mandatory as well as being a legal and ethical duty. The main contribution of this paper is to propose a new Content Authentication (CA) watermarking scheme, which aims at detecting any modification, forgery, or illegal manipulation of images even if it is small. Our proposed scheme is a fragile, secure, and a reversible watermarking scheme. It generates the watermark uniquely using a messy model. The generated watermark is embedded accumulatively; to obtain spreading over the whole image area, and embedded homogeneously; to obtain a high quality watermarked image. Our proposed scheme is a development of a recently proposed watermarking scheme. Our proposed scheme surpassed its counterpart in terms of capacity, quality, watermark spreading, fragility, and embedding time. The payload of the host image increased from 81.71 % to 93.82 %. The minimum obtained PSNR value increased from 27.15 dB to 31.76 dB. The watermark spreading percentage, or the percentage of the protected pixels, is noticeably increased. Our proposed scheme is very sensitive to modifications anywhere in the image even if it is tiny. Finally, our proposed CA scheme has a faster embedding time than that of its counterpart. We obtained an average reduction in time equals 0.15 second.[...] Read more.
DOI: https://doi.org/10.5815/ijcnis.2013.03.02, Pub. Date: 8 Mar. 2013
In this paper, we introduce a new attack, Reinforcement Swap Attack, against Directed Diffusion based WSNs, which exploits the vulnerabilities of Directed Diffusion specifications. Its main idea is the disruption of configuration information, such as routing information to misuse route establishment along the network. Our approach is to swap Directed Diffusion reinforcement rule which means that the good route is excluded and the bad route is included. Moreover, our attack is activated and deactivated periodically to prolong its lifetime and hence brings down the target network. For the proposed attack, we present analysis, simulation, and experimental measurements. We show that the system achieves maximal damage on system performance represented by many metrics.[...] Read more.
DOI: https://doi.org/10.5815/ijcnis.2012.12.02, Pub. Date: 8 Nov. 2012
The objective of this paper is to study the vulnerabilities of sensor networks, design, and implement new approaches for routing attack. As one of the cornerstones of network infrastructure, routing systems are facing more threats than ever; they are vulnerable by nature and challenging to protect.
We present a new attack, Swarm Flooding Attack, against Directed Diffusion based WSNs, which targets the consumption of sensors computational resources, such as bandwidth, disk space, or processor time. Two variants of swarm attack have been introduced: Bee and Ant. Both approaches are inspired from the natural swarming difference between bees and ants. In all cases, the strategy used to mount an attack is the same. An attack consists of a set of malicious user queries represented by interests that are inserted into the network. However, the two forms of attack vary in the synchronization aspects among attackers. These types of attacks are hard to defend against as illustrated. For each of the proposed attack models, we present analysis, simulation, and experimental measurements. We show that the system achieves maximal damage on system performance represented by many metrics.
DOI: https://doi.org/10.5815/ijcnis.2012.10.03, Pub. Date: 8 Sep. 2012
This paper is a contribution in the field of security analysis on mobile ad-hoc networks, and security requirements of applications. Limitations of the mobile nodes have been studied in order to design a secure routing protocol that thwarts different kinds of attacks. Our approach is based on the Zone Routing Protocol (ZRP); the most popular hybrid routing protocol. The importance of the proposed solution lies in the fact that it ensures security as needed by providing a comprehensive architecture of Secure Zone Routing Protocol (SZRP) based on efficient key management, secure neighbor discovery, secure routing packets, detection of malicious nodes, and preventing these nodes from destroying the network. In order to fulfill these objectives, both efficient key management and secure neighbor mechanisms have been designed to be performed prior to the functioning of the protocol.
To validate the proposed solution, we use the network simulator NS-2 to test the performance of secure protocol and compare it with the conventional zone routing protocol over different number of factors that affect the network. Our results evidently show that our secure version paragons the conventional protocol in the packet delivery ratio while it has a tolerable increase in the routing overhead and average delay. Also, security analysis proves in details that the proposed protocol is robust enough to thwart all classes of ad-hoc attacks.
DOI: https://doi.org/10.5815/ijcnis.2012.08.01, Pub. Date: 8 Aug. 2012
In this paper, a symmetric key block cipher cryptosystem is proposed, involving multiple two-dimensional chaotic maps and using 128-bits external secret key. Computer simulations indicate that the cipher has good diffusion and confusion properties with respect to the plaintext and the key. Moreover, it produces ciphertext with random distribution. The computation time is much less than previous related works. Theoretic analysis verifies its superiority to previous cryptosystems against different types of attacks.[...] Read more.
DOI: https://doi.org/10.5815/ijcnis.2012.07.03, Pub. Date: 8 Jul. 2012
In this paper, a new Chaotic Map with Block Chaining (CMBC) cryptosystem for image encryption is proposed. It is a simple block cipher based on logistic chaotic maps and cipher block chaining (CBC). The new system utilizes simplicity of implementation, high quality, and enhanced security by the combined properties of chaos and CBC cipher. Implementation of the proposed technique has been realized for experimental purposes, and tests have been carried out with detailed analysis, demonstrating its high security. Results confirm that the scheme is unbreakable with reference to many of the well-known attacks. Comparative study with other algorithms indicates the superiority of CMBC security with slight increase in encryption time.[...] Read more.
DOI: https://doi.org/10.5815/ijigsp.2012.05.05, Pub. Date: 8 Jun. 2012
In this paper, we present an efficient content based image retrieval system that uses texture and color as visual features to describe the image and its segmented regions. Our contribution is of three directions. First, we use Gabor filters to extract texture features from the whole image or arbitrary shaped regions extracted from it after segmentation. Second, to speed up retrieval, the database images are segmented and the extracted regions are clustered according to their feature vectors using Self Organizing Map (SOM). This process is performed offline before query processing; therefore to answer a query, our system does not need to search the entire database images. Third, to further increase the retrieval accuracy of our system, we combine the region features with global features to obtain a more efficient system.
The experimental evaluation of the system is based on a 1000 COREL color image database. From experimentation, it is evident that our system performs significantly better and faster compared with other existing systems. We provide a comparison between retrieval results based on features extracted from the whole image, and features extracted from image regions. The results demonstrate that a combination of global and region based approaches gives better retrieval results for almost all semantic classes.
Subscribe to receive issue release notifications and newsletters from MECS Press journals