Faculty Publications

Permanent URI for this collectionhttps://hdl.handle.net/10361/15998

Browse

Now showing 1 - 20 of 76

Open Access
Telecom customer behavior analysis using naïve bayes classifier
(Institute of Electrical and Electronics Engineers Inc., 2021-08-13) Rabiul Alam, Md. Golam; Hussain, Sajjad; Mim, Md. Mofaqkhayrul Islam; Islam, Md Tarikul
A deep understanding of customer's needs and efficient analysis of customer behavior is such a measurement tool on which the paramount success of each enterprise relies upon. The customer retention ability is being focused on effective studies of potential customers, better decision-making, and enhanced business processes. Naive Bayes (NB) classifier, a simple classifier can be used for anticipating customer behavior. Through an excellent understanding of customers and their consuming behavior, a firm can retain loyal customers, improve its online footprint, forecast how customers will react to the company's marketing strategies, enhance business policy to discover new consuming areas, and boost profits income. The goal of this study is to look into the purchasing habits of telecom firm customers to help them be one of the leading brands in terms of implementing services in the telecoms sector. We conducted our data analysis on a customer records dataset from a telecommunication company, composed of various service usage records of its customers from various states across the United States of America.
Line profile-based fingerprint matching
(© 2017 Institute of Electrical and Electronics Engineers Inc., 2/21/2017) Ali, Hafsa Moontari; Corraya, Sonia; Department of Computer Science and Engineering
Traditional fingerprint matching algorithms primarily focus on minutiae points on fingertip surface. In this paper, a novel approach is proposed for fingerprint matching that is based on ridge and valley characteristics of fingerprints. At first, the input fingerprint image is normalized and the registration point of that particular fingerprint is detected. Then a line profile is generated centering on that reference point. The distances between the reference point and ridges and the count of intersection points of line profile and ridges are stored in database. This process is repeated after every 15-degree angle to 345-degree in clock-wise direction and for orientation angle, the distances are stored sequentially. For matching intersection point count number along with the sequence of distance values are compared with the stored values. This new method can detect fingerprint from any orientation angle. Experimental result shows 90.87% accuracy of the proposed method.
Acoustic emission sensor network based fault diagnosis of induction motors using a gabor filter and multiclass support vector machines
(© 2016 Old City Publishing, Inc., 2016) Islam, Md Rashedul; Uddin, Jia; Kim, Jongmyon; Department of Computer Science and Engineering
Reliable and efficient fault diagnosis of induction motors is an important issue in industrial environments. This paper proposes a method for reliable fault diagnosis of induction motors using signal processing of acoustic emission (AE) data, including Gabor filtering and the use of multiclass support vector machines (MCSVMs), where a ZigBee based wireless sensor network (WSN) model is used for efficiently transmitting AE signals to a diagnosis server. In the proposed fault diagnosis approach, the induction motor’s different state signals are acquired through proper placement of AE sensors. The AE data are sent to a server through the wireless sensor network and decomposed using discrete wavelet transformation (DWT). An appropriate band is then selected using the maximum energy ratio, and a one-dimensional (1D) Gabor filter with various frequencies and orientation angles is applied to reduce abnormalities and extract various statistical parameters for generating features. In addition, principal component analysis (PCA) is applied to the extracted features to select the most dominant feature dimensions. Finally, one-against-one multiclass support vector machines (OAA-MCSVMs) are used to classify multiple fault types of an induction motor, where each SVM individually trains with its own features to increase the fault classification accuracy of the induction motor. In experiments, the proposed approach achieved an average classification accuracy of 99.80%, outperforming conventional fault diagnosis models.
Human development and interaction in the age of ubiquitous technology
(© 2016 IGI Global, 1/27/2016) Rahman, Hakikur; Department of Computer Science and Engineering
The human condition is affected by numerous factors in modern society. In modern times, technology is so integrated into culture that it has become necessary to perform even daily functions. Human Development and Interaction in the Age of Ubiquitous Technology is an authoritative reference source for the latest scholarly research on the widespread integration of technological innovations around the globe and examines how human-computer interaction affects various aspects of people's lives. Featuring emergent research from theoretical perspectives and case studies, this book is ideally designed for professionals, students, practitioners, and academicians.
Open innovation: reaching out to the grass roots through SMEs - exploring issues of opportunities and challenges to reach economic sustainability
(© 2016 IGI Global, 6/27/2016) Rahman, Hakikur; Department of Computer Science and Engineering
While talking about successful entrepreneurship and value addition within an enterprise through innovation, one could realize that the innovation paradigm has been shifted from simple introduction of new ideas and products to accumulation of diversified actions, actors and agents along the process. Furthermore, when the innovation process is not being restricted within the closed nature of it, the process takes many forms during its evolution. Innovations have been seen as closed innovation or open innovation, depending on its nature of action, but contemporary world may have seen many forms of innovation, such as technological innovation, products/service innovation, process/production innovation, operational/management/organizational innovation, business model innovation or disruptive innovation, though often they are strongly interrelated. Definition of innovation has also adopted many transformations along the path, incorporating innovations within the products, process or service of an enterprise to organizational, marketing, or external entities and relations. Nature and scope of agents and actors even varies widely within the innovation dynamics, when the open innovation techniques are being applied to enterprises, designated as the small and medium enterprises (SMEs). Researching in this paradigm, one has to look for some underlying issues that should be attended through responding to research questions as the research continues. Among many of the fundamental questions on innovation advancement for SMEs development there are a few, how to acquire precise information on the flow-chart of their business operations, gain knowledge on specific parameters of their business processes, utilize existing potential capacities to extend their knowledge towards successful innovation acquisition and dissemination, and extend their knowledge platform through various capacity development initiatives. They aggregate further, when issues of opportunities and challenges are being researched along the path of SME development through open innovation. Rationale of this research is to ascertain diverse aspects of opportunities and challenges surrounding the open innovation processes, and design action plans to empower SMEs in reaching out to the grass roots communities utilizing open innovation strategies. Primary focus of this research is to enable SMEs in finding out their innovation potentiality and empower them through various capacity development initiatives. However, the specific focus will adhere to adaptable technology transfer through open innovation. Along the route to justify the research potential and validate the research hypotheses (whether this research will add any economic value or knowledge gain), this study will conduct extensive literature review on various patterns of open innovation (crowdsourcing or collaborative), investigate case studies to learn about intricate issues surrounding their operational strategies (conducted by European Commission, OECD and similar institutions) and conduct surveys among selected SMEs (email, web based, egroups) in several phases. Research design includes formulation of strategies to resolve acquired research questions; collection and recording of the evidences obtained from the literature review or case studies or surveys; processing and analyzing gathered data and their appropriate interpretations; and publication of results. Analysis will include both qualitative (descriptive and exploratory) and quantitative (inferential statistics) methods.
Intelligent intrusion prevention system for households based on system-on-chip computer
(© 2016 IEEE, 2016-05) Rashid, Md Tahmid; Abir, Imtiaz Kalam; Shourove, Niaz Sharif; Muntaha, Rakibun; Rhaman, Md Khalilur; Department of Computer Science and Engineering
Over the course of the last few decades, security systems have undergone radical overhauls and have transitioned into such sophisticated devices that some even completely shun any form of human interventions. In spite of its tremendous innovations, however, such setups have not yet made their way into homes and residences, partly owing to the fact that the equipment bear high price tags and lack user friendliness. This paper propositions a unique security system for residences that aims to solve the shortcomings of commercial alternatives. Basically, a Raspberry Pi 2 System-on-Chip (SoC) computer equipped with a camera module and placed outside the entrance of a house would attempt to identify would-be visitors by the means of image processing and voice identification. The system would spontaneously greet known faces with a friendly voice greeting and unlock the main door to allow them to enter the home. In case of unknown faces, the system would initiate a video call to the home's owner through a smartphone application and allow the owner to communicate with the unknown visitor. Upon receiving approval from the owner, the visitor's details would be entered into the system and the main door would be unlocked. If the owner denies access, he or she has the option to verbally ask the unknown visitor to leave or in extreme cases, press a button to call the police.
A study on image enhancement methods for printed Bangla document images
(© 2016 IEEE, 2016) Islam, Anika Binte; Chowdhury, Fahim Salam; Nusrat, Fariha; Kabir, Kazi Lutful; Sarwar, Hasan; Department of Computer Science and Engineering
Image enhancement is the process of modification of digital images in order to make quality resultant images such that those can be subjected to further manipulation (like further image analysis). There are several ways to implement image enhancement techniques. In this paper, we have studied a number of image enhancement methods consisting of binarization techniques and hybridization of such techniques. After applying those methods on scanned Bangla document images, we have analyzed the reconstructed images. Then, we have gone for the quantitative analysis portion. To accomplish that, we have measured the recognition rate and then examined the improvement ratio of the unprocessed samples versus reconstructed images in terms of recognition rate. Our analysis critically emphasizes the requirement to apply image enhancement methods. This study will certainly help the researchers to select appropriate image enhancement method(s) for processing low quality Bangla printed document images.
Reinforcement learning based autonomic virtual machine management in clouds
(© 2016 IEEE, 11/28/2016) Habib, Arafat; Khan, Muhidulislam; Department of Computer Science and Engineering
Cloud computing is a rapidly emerging field, services and applications are more or less 24/7. Resource dimensioning in this field is a great issue. Research is already going on to imply reinforcement learning to automate decision making process in case of addition, reduction, migration and maintenance of the Virtual Machines (VM) to balance the service level performance and VM management cost. Models have been proposed in this case based on Q-learning, a very popular reinforcement learning technique that is used to find optimal action selection policy for any finite Markov Decision Process (MDP). In this paper we propose to work with the challenges like proper initialization of the early stages, designing the states, actions, transitions using Markov Decision Process (MDP) and solving the MDP with two popular reinforcement learning techniques, Q-learning and SARSA(Λ).
Viewing angle enhancement of a real-time integral imaging system using multi-directional projections and GPU parallel processing
(© 2017 SPIE, 2017-02) Alam, Md Ashraful; Islam, Md Sifatul; Tareque, Mohd Zishan; Subhani, Mahfuze; Rafi, Mohammad Rashidur Rahman; Alam, Md Ashraful; Kim, Nam; Department of Computer Science and Engineering
A novel method of viewing angle enhancement of a real-time integral imaging system using multi-directional projections and GPU parallel processing is proposed. The proposed system is composed of three processes: information acquisition of real objects, generation of multi-directional elemental image sets, and reconstruction of 3D images by using multidirectional projections scheme. To implement this system, depth and color (RGB) information of each object point are captured by a depth camera; then, a dynamic algorithm and GPU parallel processing are used for generating multidirectional elemental image sets to be illuminated in different directions as well as to maintain a real-time processing seed; and finally, 3D images are reconstructed by using a time-multiplexed multi-directional projection scheme through an appropriate optical setup of a projection-type integral image system. Multi-directional illuminations of elemental image sets enhance the optical ray divergence of reconstructed 3D images according to the directional projection angles. Hence, a real-time integral imaging system with enhanced viewing angle is achieved.
A web and software-based approach blending social networks for online Qur’anic arabic learning
(© 2017 Zarka Private Univ., 2017-01) Abdullah, Matin Saad; Pathan, Al Sakib Khan; Al-Shaikhli, Imad Fakhri Taha; Department of Computer Science and Engineering
About 80 percent of the world’s Muslim populations are non-native speakers of the Arabic language. Since it is obligatory for all Muslims to recite Qur’an in Arabic during prayers, an extraordinary social phenomenon has taken place in some parts of the Muslim world: Muslims are taught the complex phonological rules of the Arabic language in the context of Qur’an and they recite the “sounds” of Qur’an often understanding very little. This has given rise to a demographic segment of adult learners whose main learning goal is recalling a closed set of syntactic rules and vocabularies in the context of Qur’an while reciting or listening to it so that they can reconstruct a meaning in their native-language. Despite the availability of some resources for learning language for this specific purpose, according to our detailed investigation, no work has explored the possibilities of emerging adaptive and intelligent systems for collaborative learning to address this challenge. The goals of this work are: To determine the applicability of learner corpus research, declarative memory modelling, and social learning motivation on the learners’ specific pedagogical objectives and to use the Design-Based Research methodology (DBR) to optimize the design of such a system in real-life setting to observe how the different variables and elements work out. We present here, a prototype to gather requirement analysis of such a system by bootstrapping a user community. The compiled data were used to design an initial architecture of an intelligent and adaptive Qur’anic Arabic learning system.
A novel parallel texture feature extraction method using log-gabor filter and singular value decomposition (SVD)
(© 2017 ACM, 2017-01) Ratul, Md Aminur Rab; Raja, Sharif Ahmmad; Uddin, Jia; Department of Computer Science and Engineering
Texture feature extraction consolidated with texture feature detection and feature matching solves many typical problems of image processing and computer vision; such as, texture classification, pattern recognition, object detection, and image segmentation. Through this paper, a new method for texture feature extraction is presented which uses Log-Gabor Filter and Singular Value Decomposition (SVD) algorithm. In the proposed model, sample images are converted to gray level images. And then, to elicit suitable distinctive texture orientation, a 2D Log- Gabor filter with various frequencies and different edges disintegrated with the SVD employ on each converted gray level images. Finally, singular values of SVD used as feature vector for this texture feature extraction model. For training and testing of experimental datasets, Naive Bayes classifier has been used. The Log-Gabor and SVD based feature extraction shows improved performance by exhibiting higher classification accuracy for our tested dataset compare to conventional Gabor and SVD feature extraction method. Furthermore, in order to decrease the computational and time complexity, an NVIDIA GeForce GTX780 GPU is used to implement our proposed model in parallel. The GPU implementation of proposed model showed average 3X speedup for per image than conventional CPU implementation.
An improved chromosome sorting algorithm by permutation group-based inverted block-interchanges
(© 2016 IEEE, 2016-12) Tabassum, Afrida; Abdullah, Wali Mohammad; Manjur, Najia; Islam, Rehennuma; Ahmed, Ashfaque Uddin; Morshed, Shourav Imtiaz; Tinni, Rizia Iqbal; Kabir, Kazi Lutful; Islam, Muhammad Nazrul; Department of Computer Science and Engineering
In comparative genomics, a common issue is to find the shortest course of operations for the transformation of one genome into another. During rejoining the blocks, inverted block interchange, a more distinct form of block interchange, gives the option to the blocks to reverse their orientation. In this particular paper, we have presented an improved algorithm to sort circular chromosomes by inverted block interchanges and also compared the performances with a previous algorithm. Our experimental findings justify the fact that accommodation of inverted block interchange ensures results with equal or less number of series of operations in terms of block interchanges.
PNGP : a social relationship based routing algorithm for pocket switched network
(© 2016 IEEE, 2016-12) Barua, Rupa; Shadman, Shauvik; Chakrabarty, Amitabha; Department of Computer Science and Engineering
Transferring messages or packets through a network using human mobility as a means, requires intelligent routing decisions. Throughout an opportunistic network, nodes travel with a few messages and searches for an opportunity to deliver the message. A good delivery rate of such messages in these sort of networks is a challenge which might require complex calculations and several intricate decision making processes. With a view to improve efficiency and to deliver a simpler way of routing, this paper will present a new routing algorithm called Popular Node Gateway Protocol(PNGP). This new algorithm is tested against 3 other algorithms using our own built simulator and proves to be better in terms of delivery ratio delay and transmission cost. Unlike other algorithms this algorithm provides a simpler way of routing thus making this faster and much more efficient.
A novel modified SFTA approach for feature extraction
(© 2016 IEEE, 9/22/2016) Hasan, Md Junayed; Uddin, Jia; Pinku, Subroto Nag; Department of Computer Science and Engineering
To increase the efficiency of conventional Segmentation Based Fractal Texture Analysis (SFTA), we propose a new approach on SFTA algorithm. We use an optimum multilevel thresholding hybrid method of Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), called HGAPSO with the optimization technique for classification based on grey level range to get more accurate output. Experimental results show that proposed approach exhibits average 2% higher classification accuracy than conventional SFTA for our tested dataset.
Efficient and fast convergent consensus algorithms for faulty nodes tracking in distributed wireless sensor networks
(© 2016 IEEE, 2016-09) Khan, Muhidulislam; Hossain, Rajkin; Department of Computer Science and Engineering
One of the challenging issues in a distributed computing system is to reach on a decision with the presence of so many faulty nodes. These faulty nodes may update the wrong information, provide misleading results and may be nodes with the depleted battery power. Consensus algorithms help to reach on a decision even with the faulty nodes. Every correct node decides some values by a consensus algorithm. If all correct nodes propose the same value then all the nodes decide on that. Every correct nodes must agree on the same value. Faulty nodes do not reach on the decision that correct nodes agreed on. Binary consensus algorithm and average consensus algorithm are the most widely used consensus algorithm in a distributed system. We apply binary consensus and average consensus algorithm in a distributed sensor network with the presence of some faulty nodes. We evaluate these algorithms for better convergence rate and error rate.
An adaptive digital image watermarking scheme with PSO, DWT and XFCM
(© 2017 IEEE, 2017-02) Mitashe, Mashruha Raquib; Habib, Ahnaf Rafid Bin; Razzaque, Anindita; Tanima, Ismat Ara; Uddin, Jia; Department of Computer Science and Engineering
In this paper, a novel adaptive digital image watermarking model based on modified Fuzzy C-means clustering is proposed. For watermark embedding process, we used Discrete Wavelet Transform (DWT). A segmentation technique XieBeni integrated Fuzzy C-means clustering (XFCM) is used to identify the segments of original image to expose suitable locations for embedding watermark. We also pre-processed the host image using Particle Swarm Optimization (PSO) to lend a hand to the clustering process. The goal is to focus on proper segmentation of the image so that the embedded watermark can withstand common image processing attacks and provide security to digital images. Several attacks were performed on the watermarked images and original watermark was extracted. Performance measures like PSNR, MSE, CC were computed to test the extracted watermarks with and without attacks. Experimental results show that the proposed scheme has performed well in terms of imperceptibility and robustness when compared to other watermarking models.
Sensing properties of gas sensor based on adsorption of NO2 with defect, pristine, Fe and Si-MoS2 layer
(© 2014 Sumy State University, 2014) Shakil, Shifur Rahman; Khan, Sabbir Ahmed; Department of Computer Science and Engineering
Two-dimensional (2D) layered materials are currently being considered as entrant for future electronic devices. Molybdenum disulphide (MoS2) belongs to a family of layered transitional metal dichalcogenides(TMDS),has a unique characteristics of showing intrinsic semiconducting nature is being considered a major advantageous over graphene (which has no intrinsic band gap) as a two-dimensional (2D) channel material in field effect transistors(FET). In the paper, the results of investigations are presented concerning the affects of adsorption of NO2 gas on the surface of MoS2, defect-MoS2, Si-MoS2 and Fe-MoS2 layer. The changes density of states (DOS) and electrostatic difference potential of Si-MoS2 by applying different gate voltage were studied. We proposed that, NO2 might play an important role on MoS2 layer that can be used as gas sensor. In the research, it has been shown that in the case of gas sensor, the adsorption of NO2 with MoS2, Fe-MoS2, Si-MoS2 and defect-MoS2 play an important rule for sensing behavior.
Humidity sensor using surface adsorbed channel modulated Graphene nanoribbon: NEGF approach
(© 2014 Institute of Electrical and Electronics Engineers Inc., 2015-01) Shakil, Shifur Rahman; Zohra, Fatema Tuz; Pramanik, Parna; Tushar, Raihanul Islam; Department of Computer Science and Engineering
Electronic transport properties of Graphene nanoribbon (GNR) are explored by self-consistent numerical simulation. The effects of vapour (H2O) adsorption in GNR sheets on its current-voltage (I-V) characteristics, conductivity, electrostatic difference potential and device density of states were studied with necessary physical insight. We found that conductivity of GNR sheet increases in some particular energy range with the increase in number of H2O molecules adsorbed. Then we encounter the effects of metal-GNR contact to investigate the practical performance of GNR as a medium of humidity sensing. Later we propose a simple schematic model for GNR Humidity Sensor (GNRHS) to investigate its performance in applied regime.
Propasal of possible OTEC sites in Bangladesh
(© 2014 IEEE Computer Society, 2013) Shakil, Shifur Rahman; Hossain, Md. Safwat; Rouf, Nirjhor Tahmidur; Department of Computer Science and Engineering
Bangladesh, a new name among the fast developing country, has a huge population of more than 160 Million people and a large demand for electricity. Although Bangladesh is still struggling to provide a stable energy source to the growing industrial sectors and general mass, the geographical location of this country is very advantageous for the development of renewable energy. The site of Bangladesh is naturally gifted area to establish alternate energy collection sources like Ocean Thermal Energy Conversion (OTEC). Since Bangladesh lies just beneath the tropic of cancer and on the shore of the Bay of Bengal, the two vital elements: constant sunlight and large littoral areas needed for OTEC can easily be found in this region. With huge water bodies throughout, this zone is very much suitable for developing multiple OTEC plant size.
Performing sentiment analysis in Bangla microblog posts
(© 2014 IEEE Computer Society, 2014) Chowdhury, Shaika; Chowdhury, Wasifa; Department of Computer Science and Engineering
Much of the research work on sentiment analysis has been carried out in the English language, but work in Bangla is limited to only news corpus and blogs. Microblogging sites are becoming a valuable source for publishing huge volumes of user-generated information, as users express their views, opinions, and sentiments over various topics. In this paper, we aim to automatically extract the sentiments or opinions conveyed by users from Bangla microblog posts and then identify the overall polarity of texts as either negative or positive. We use a semi-supervised bootstrapping approach for the development of the training corpus which avoids the need for labor intensive manual annotation. For classification, we use Support Vector Machine (SVM) and Maximum Entropy (MaxEnt) and do a comparative analysis on the performance of these two machine learning algorithms by experimenting with a combination of various sets of features.

Browse

Recent Submissions