Accepted Papers

  • Coding Algorithm Optimization and DSP Implementation of Selectable Mode Vocoder
    Li Qiang, Zhangling, Chongqing University of Posts and Telecommunications,china

    ABSTRACT

    Based on the coding principle of the selectable mode vocoder, a low bit variable rate vocoder is realized by adjusting the structure of algorithm and selecting the encoding rate. In order to overcome the shortcomings of large coding computational complexity due to the iterative algorithm performing a full search for the code-book, an improved code-book search algorithm is proposed. According to the frame classification information and the coded frame characteristic parameters,the selection criterion is changed to reduce the number of search sub code-book, cut down the complexity of code-book search. The vocoder is implemented on the TMS320C6713 DSK, the complexity of the code-book search module and the encoding module is evaluated. Subjective and objective quality test on the vocoder results indicate that the synthesized speech of the vocoder still has high intelligibility in the case of greatly reducing the complexity of the code-book search.

    KEYWORDS

    Selectable mode vocoder, Fixed codebook search, Characteristic parameters & DSP implementation

    A High Timeslot Utilization Data Transmission Mechanism for Terahertz Inter-PAN
    Jieli Tian, Zhi Ren and Yhui Lv, Chongqing University of Posts and Telecommunications, China

    ABSTRACT

    Inter-PAN communication for terahertz wireless personal area network requires bridge node to forward control messages. The existing inter-PAN communication solutions have some problems such as a waste of timeslots due to unbalanced network traffic and large control overhead caused by the unified superframe mechanism. Aiming at these problems, a high timeslot utilization inter-PAN data transmission mechanism for terahertz wireless personal area network is proposed. Simulation results show that the protocol can effectively use the timeslot and increase network throughput.

    KEYWORDS

    terahertz; inter-PAN; coordinated superframe; timeslot utilization

    Available Bandwidth Estimation Algorithm Based On Channel Monitor
    Yu Xiang, Meng Yanqun , Wang Shiyan and Liu Li, Chongqing University of Posts and Telecommunications, China

    ABSTRACT

    The available bandwidth estimation of the link is of great significance to multimedia transmission.There are still some problems that can't be solved effectively according to the current bandwidth estimation methods, such as bandwidth overhead, node synchronization and packet collision. This thesis presents a passive bandwidth estimation method based on monitoring channel. The algorithm analyse the collision back-off, transceiver node monitor channel idle synchronously and how data packet collision impact network bandwidth, considering the control information channel duty cycle, so as to accurately estimate the link available bandwidth. Finally, the simulation results of NS2 show that the algorithm has high accuracy.

    KEYWORDS

    Link available bandwidth estimation; synchronous idle probability; packet collision; control information

    Making MDD Agile the Agile Model-Driven Method
    Klaus Mairon1, Martin Buchheit1, Martin Knah1and Shirley Atkinson2, 1Hochschule Furtwangen University, Germany and 2Plymouth University, United Kingdom.

    ABSTRACT

    This article takes up the idea of model-driven development in a new way and analyses existing points of criticism of this approach, which is well established in practice. The advantages of model-driven development seem obvious on the one hand, on the other hand there is criticism of the practicable use and the accusation of missing suitable process models. This environment of professional software development is currently characterized by the use of agile process models such as Scrum, XP, etc. However, an agile process model for the use of model-driven development (MDD) does not yet exist. An analysis of the similarities between existing approaches to MDD process models and existing agile modelling techniques forms the basis for the definition of a new agile process model. The Agile Model-Driven Method (AMDM) is the result of these studies.

    KEYWORDS

    Model-Driven Development, Model-Driven Architecture, Process Model, Agile Method, Software Engineering, UML

    NUMA Awareness:Improving Thread and Memory Management
    Maria Patrou, Kenneth B. Kent, Gerhard W. Dueck, Charlie Gracie and Aleksandar Micic

    ABSTRACT

    Speci c characteristics of hardware can a ect the performance of applications. Hardware awareness becomes essential during runtime for both memory allocation and access and threads' processor locations. Non-uniform Memory Access (NUMA) systems use di erent types of memory accesses and depend on several hardware resources in speci c topologies. IBM's Java Virtual Machine (JVM) examines the underlying hardware and identi es a NUMA architecture, while using memory and threads from the available nodes in a distributed manner. A design for a nodeisolated memory and thread policy is proposed, called NumaVM. A node-heap resize functionality to retrieve memory, based mostly on each node's memory information, is further described. Finally, di erent modes regarding hardware and thread characteristics are used in order to nd an optimal one and identify the application attributes that can bene t from speci c modes, based on the underlying hardware.

    KEYWORDS

    NUMA, Non-uniform memory access, thread anity, memory management, garbage collection

    Mobile Application Development Using Gamification Techniques To Simulate The Admission Exam To The National University Of Colombia
    Michael Valencia Ramirez Universidad Distrital Francisco Jose de Caldas Bogota, Colombia

    ABSTRACT

    This article aims to show the advances in the development of a mobile application to simulate the admission exam to the National University of Colombia (UN), which makes use of gamification techniques. For young graduates of high school, one of the main concerns is to apply to the university and be admitted. The National University of Colombia, because it is public in nature and has a high academic level, it is in great demand in the country, each semester there are approximately 72 thousand applicants and only about 10% is accepted due to the low availability of university quotas. The selection process is carried out by means of an admission examination, which evaluates the skills in reading comprehension and logical-mathematical reasoning of the aspirant. Faced with this problem and with the unstoppable evolution of technology, it is vital to integrate the ICT tools (Information and Communication Technologies), as components in learning and preparing them, a mobile application is developed, with the aim of helping aspirants to have a better preparation for the exam. It is taken as population study of 20 students of the eleventh grade belonging to the Liceo Cultural Ernesto Guhl School in Bogota, Colombia.

    KEYWORDS

    Education, Gamification, University, ICT, Mobile application.

    A silent hack detection based on deep-learning technique
    SalehAlbahli,IT Department, College of Computers, Qassim University, KSA

    ABSTRACT

    Sharing information has been democratized with the rise of social networks. Consequently, increasing the usage of Social Network, especially Twitter platform, leads to growing malicious activities. With a silent hack, a hacker can continuously dig around to control over victim's account. In this paper, an observed direct impact to users' security and privacy has been identified. Therefore, we address hidden tactics in the problem-specific feature engineering with detailed results to show how deep leaning classifiers are promising direction to understand sentiment than classical machine learning. Thus, we focus on the state of the art Deep learning techniques by constructing a model to detect behavioral changes of users. Our evaluation shows that working with just classical machine algorithms to analyse social data do not achieve higher performance than deep learning algorithms. This will open directions for using deep learning for similar problems. Moreover, our results demonstrate the shortages of classical Machine Learning classifiers compared to Deep learning and how they can be mitigated.

    KEYWORDS

    Deep learning, Machine learning, silent hacking, social data,behaviors, analysis, Twitter.

    Deep learning: Building an Image Retrieval System for a fashion ecommerce company
    Rui Machado1, Joao Gama11Faculty of Economics, University of Porto

    ABSTRACT

    Deep learning is a very trendy topic now, showing high accuracy in image based systems that can go from image segmentation to object detection and image retrieval. Because of this, multiple researchers and companies have been building and sharing work in the community, including pre-trained convolutional neural networks, available for public use. This work follows the trend and delivers an experimental study using deep learning for building a visually similar image retrieval application, comparing three different convolutional neural architectures for feature extraction and six distance indexes for similarity calculation in a real-world image retrieval problem, using real data from a fashion e-commerce platform from Morocco. The main objectives of the proposed methodology were to answer three fundamental questions, regarding how can deep learning be applied to the construction of a visual similarity based image retrieval system for a fashion ecommerce platform, what similarity measures best fit this problem for image retrieval and assess if using deep learning for image retrieval is already an enterprise option, available to any fashion ecommerce company or still a research branch. After testing all the different combinations, we can conclude that for this dataset, Vgg19 combined with a correlation coefficient for similarity calculation is the tuple that best maximizes the similarity between a search image and its retrieved neighbors.

    KEYWORDS

    Deep learning; Image-based search; convolutional Neural networks; fashion image retrieval

    Soccer Event Detection
    Abdullah Khan12, Beatrice Lazzerini2, Gaetano Calabrese3, and Luciano Sera ni3
    1Department of Information Engineering, University of Pisa, Pisa, Italy
    2 Department of Information Engineering, University of Florence, Florence, Italy
    3Fondazione Bruno Kessler, Trento, Italy

    ABSTRACT

    The research community is interested in developing automatic systems for the detection of events in video. This is particularly important in the eld of sports data analytics. This paper presents an approach for identifying major complex events in soccer videos, starting from object detection and spatial relations between objects. The proposed framework, rstly, detects objects from each single video frame providing a set of candidate objects with associated con dence scores.The event detection system, then, detects events by means of rules which are based on temporal and logical combinations of the detected objects and their relative distances. The e ectiveness of the framework is preliminary demonstrated over di erent events like \Ball possession" and \Kicking the ball".

    Keywords

    Event detection in video, simple events, complex events.

    Machine learning systems based on xgBoost and MLP neural network applied in satellite lithium-ion battery sets impedance estimation
    Thiago Henrique Rizzi Donato, Department of Applied Computation,National Space Research Institute - INPE,Sao Jose dos Campos, SP 12227-010, Brazil Marcos Quiles,Division of Machine Learning,Federal University of Sao Paulo - UNIFESP,Sao Jose dos Campos, SP 12231-280, Brazil

    ABSTRACT

    In this work, the internal impedance of the lithium-ion battery pack (important measure of the degradation level of the batteries) is estimated by means of machine learning systems based on supervised learning techniques MLP - Multi Layer Perceptron - neural network and xgBoost - Gradient Tree Boosting. Therefore, characteristics of the electric power system, in which the battery pack is inserted, are extracted and used in the construction of supervised models through the application of two di erent techniques based on trees and neural network. Finally, with the application of statistical validation techniques, the accuracy of both models are calculated and used for the comparison between them and the feasibility analysis regarding the use of such models in real systems.

    Keywords

    Lithium-ion battery, Internal impedance, State of charge, Multi layer per-ceptron, Gradient tree boosting, xgBoost

    Twitter Sentiment Analysis of New IKEA Stores Using Machine Learning
    Yujiao Li and Hasan Fleyeh, Dalarna University,Borlange, Sweden

    ABSTRACT

    The aim of this paper is to study public emotion and opinion concerning the opening of new IKEA stores.Specifically, how much attention are attracted, how much positive and negative emotion are aroused, what IKEA related topics are talked due to this event. Emotion is difficult to measure in retail due to data availability and limited quantitative tools. Twitter texts, written by the public to express their opinion concerningthis event, are used as a suitable data source to implement sentiment analysis. Around IKEA opening days, some local people post IKEA related tweets to express their emotion and opinions on that. Such "IKEA" contained tweets are collected for opinion mining in this work. To compute sentiment polarity of tweets, lexicon-based approach is used for English tweets, and machine learning methods for Swedish tweets. The conclusion is new IKEA store are paid much attention indicated by significant increasing tweets frequency, most of them are positive emotions, and four studied citizens have different topics and interests related IKEA. This paper extends knowledge of consumption emotion studies of pre-purchase, provide empirical analysis of IKEA entry effect on emotion. Moreover, it develops a Swedish sentiment prediction model, elastic net method, to compute Swedish tweets' sentiment polarity which has been rarely conducted.

    KEYWORDS

    Big-box effect, Opinion analysis, customer emotion, elastic net model, text mining


    The Prediction Of Student Failureusing Classification Methods:A Casestudy
    Mashael Al luhaybi, Allan Tucker , Leila Yousefi, Computer Science Department, Brunel University, London, UK

    ABSTRACT

    In the globalised education sector, predicting student performance has become a central issue for data mining and machine learning researchers where numerous aspects influence the predictive models. This paper attempts to apply classification algorithms to evaluate student' sperformancein the higher education sector and identify the key features affecting the prediction process based on a combination of three major attributes categories. These are: admission information, module-related data and 1st year final grades. For this purpose, J48 (C4.5) decision tree and Naive Bayes classification algorithms are applied on computer science level 2studentdatasets at Brunel University London for the academic year 2015/16. The outcome of the predictive model identifies the low, medium and high risk of failure of students. This prediction will help instructors to assisthigh-risk students by making appropriate interventions.

    KEYWORDS

    Prediction, classification, decision tree, Naive Bayes, student performance.

    On The Reselection Of Seed Nodes In Independent Cascade Based Influence Maximization
    Ali Vardasbi, Heshaam Faili1, Masoud Asadpour,School of Electrical and Computer Engineering, College of Engineering, University of Tehran, Tehran, Iran

    ABSTRACT

    Influence maximization serves as the main goal of a variety of social network activities such as viral marketing. The independent cascade model forthe influence spread assumes a one-time chance for each activated node to influence its neighbors. This reasonable assumption cannot be bypassed, since otherwise the influence probabilities of the nodes would be altered. On the other hand, the manually activated seed set nodes can be reselected without violating the model parameters or assumptions. The reselection of a seed set node, simply means paying extra budget to a previously paid node in order for it to retry its influential skills on its uninfluenced neighbors. This view divides the influence maximization process into two cases: the simple case where the reselection of the nodes is not considered and the reselection case. In this study we will analyze real world networks in the reselection case. First we will show that the difference between the simple and the reselection cases constitutes a wide spectrum of networks ranging from the reselection-free to the reselection-friendly ones. Then we will experimentally show a significant entanglement between this and influence spread dynamics as well as other structural parameters of the network.

    Keywords

    Influence Maximization; Network Structure; Independent Cascade; Maximization over Integer Lattice; Core Decomposition

    Deep Learning Based Data Governance for Chinese Electronic Health Record Analysis
    Junmei Zhong1, Xiu Yi2, Jian Wang2, Zhuquan Shao2, Panpan Wang2, and Sen Lin2, 1Inspur USA Inc 2010 156th Ave NE Bellevue, WA 98052,2Inspur Software Group, Technology Center 1036 Langchao Rd., Jinan, China.

    ABSTRACT
    Electronic health record (EHR) analysis can leverage great insights for improving the quality of human health care. However, the low data quality problems of missing values, inconsistency, and errors in the data columns hinder buildingrobust machine learning models for data analysis. In this paper, we develop a methodology ofartificial intelligence (AI)-based data governance to predict the missing values or verify if the existing values are correct and what they should be when they are wrong. We demonstrate the performance of this methodology through a case study ofpatient gender prediction and verification. Experimental resultsshow that the deep learning algorithm works very wellaccording to the testing performance measured by the quantitative metric of F1-Score, and it outperformssupport vector machine (SVM) models with different vector representations for documents.
    Keywords

    EHR Analysis, Data Governance, Vector Space Model, Word Embeddings, Machine Learning, Convolutional Neural Networks.

    Single Image Dehazing with Lab Analysis
    Jehoiada K Jackson1, She Kun1, and Raphea1 Akande2 1School of Information and Software Engineering, University of Electronic Science and Technology China 2Department of Computer Science, University of Texas at El Paso

    ABSTRACT

    Images acquired by visual framework are genuinely corrupted under cloudy and foggy climate, therefore affecting detection, tracking and recognition of images. Thus, restoring the true scene from a hazy image is of great significance. To solve this problem, this paper presents a real time effective dehazing algorithm for hazy surveillance images. This algorithm is based on Histogram and a filtering manipulation on La*b* color channel. In the proposed algorithm, the input RGB night image is transformed into La*b* color channel then, contrast limited adaptive histogram equalization(CLAHE) and a smoothing operation is applied respectively and simultaneously on the luminosity layer "L" and the two color channels (a* and b*)of the La*b* color space. The channels are merged back to obtain a new enhanced image, which is transformed back to RGB image. Experimental results show the effectiveness and the short computational time of the proposed algorithm.

    Keywords

    CLAHE, image dehazing, histogram, median filter.

    Conditional Random Field Model For Blood Vessel Segmentation In Fundus Images
    Guo Ying and Yang Yuhui Shenyang University of Technology , School of Information Engineering , Shenyang, China.

    ABSTRACT

    In order to better diagnose eyes diseases, this paper presents a method for blood vessel segmentation based on conditional random field model(CRF). This method calculates the interaction between the long-distance pixels in the image and effectively enhances the detection ability and connectivity of slender structures. In this paper, the original method used for large area segmentation is successfully applied to the blood vessels segmentation in the fundus image. At the same time, Structured Output Support Vector Machine (SOSVM) is used in this paper to realize the automatic adjustment of the parameters involved in the proposed method. The results tested of DRIVE and HRF database show that the segmentation method of this paper can improve the problems of poor continuity, vascular fusion and rupture in thinner vessels. Compared with some existing methods, this proposed method has better improvement in sensitivity and specificity.

    Keywords

    Blood vessel segmentation, Fundus imaging, Conditional Random Fields, Structured Output SVM

    A Dynamic Gesture Recognition Based On Sparse Representation
    Ying Guo and Siman Bi,College of Information Science and Engineering, Shenyang University of Technology,, Shenyang, China.

    ABSTRACT

    Aiming at improving the accuracy and robustness of dynamic gesture recognition, this paper develops a method based on gesture trajectory and sparse representation. This method classifies the tracked gesture directly which omits the feature extraction in many traditional methods, the classification is transformed into sparse representation overcomplete dictionary and gesture category can be added or deleted as demanded. This method uses a class-based dictionary learning method to obtain an optimized overcomplete dictionary to reduce the computing cost and recognition time. Consider that different gestures -or even different instances of the same gesture- generally have different amplitude and speed, which brings difficulty for recognition, we add a unique category label for each type of dynamic gesture and closely associates the label information with the atomic items of the dictionary through a predefined discriminative sparse coding so that dynamic gestures of the same class have similar sparse codes and those in different classes have dissimilar sparse codes, which strengthens the classification effect of the gesture recognition. Experimental results on database of self-defined 10 types of dynamic gesture set validate the availability of the proposed recognition method.

    Keywords

    Dynamic Gesture Recognition, Sparse Representation, Dictionary Learning

    Empirical Comparison Of Visual Descriptors For Ulcer Recognition In Wireless Capsule Endoscopy Video.
    Ouiem Bchir1, Mohamed Maher Ben Ismail1, and Nourah AL_Aseem1,2, 1Computer Science Department, College of Computer and Information Sciences,King Saud University,1,2Computer Science Department, College of Engineering and ComputerSciences,Prince Sattam Bin Abdulaziz University.

    ABSTRACT

    In this work, we empirically compare the performance of various visual descriptors for ulcer detection using real Wireless Capsule Endoscopy WCE video frames. This comparison is intended to determine which visual descriptor represents better WCE frames, and yields more accurate gastrointestinal ulcer detection. The extracted visual descriptors are fed to the ulcer recognition systemwhich relies on Support Vector Machine (SVM) classification to categorize WCE frames as "ulcer" or "non-ulcer".

    Keywords

    Visual descriptors, Ulcer detection, Wireless Capsule Endoscopy

    Iterative HAAR-DWT Based Efficient Image Steganography
    Aditi Singh1, K S Venkatesh2 and Vikas Patidar3Department of Electrical Engineering, Indian Institute of Technology Kanpur, Kanpur-208016, Uttar Pradesh, India.

    ABSTRACT
    In image steganography, the transfer domain provides better concealment of the secret image in the cover image, and has therefore proved much more reliable than spatial domain. In this paper, we attempt to maximize the retrieved secret PSNR against the original secret, while simultaneously minimizing the cover image degradation. This paper is built upon Discrete-Wavelet Transform to process the image while the Least Significant Bit method to store the information. We follow a principle of priority ordering the wavelet subspaces of both the secret and the cover with a view to make for the most efficient concealment. We propose the product of the secret and cover image PSNR and SSIM measures as the quantities to be maximized as it provides a more comprehensive evaluation of system performance, and study the performance against the choice of the number of levels of wavelet decompositions.
    Keywords

    Cover; Secret; Stego; Embed; HAAR-DWT; LSB

    Non-Blocking Queue For Light Data In RTSJ
    Emad Aloqayli1 and Ahmad Habboush21Department of Software Engineering, Jadara University, Irbid, Jordan 2supDepartment of Computer Science, Jerash University, Jerash, Jordan

    ABSTRACT

    Queue data structure are in the core of most of the applications. Real Time Specifications for Java (RTSJ) introduced to enable developing realtime applications in Java. However, RTSJ does not include any general purpose implementation of the queue data structure. In this paper, a general purpose queue implementation is proposed to enable communicating small size data sizes between the different threads accessing the data structure concurrently. The results show that the proposed implementation achieve promising results compared to a previously proposed implementation in [1], and a locked-based version.

    Keywords

    Algorithms, Queue, Data structure, RTSJ, Java

    Data Retention In European Union; The Merger Of Big Data, Telecommunication Analytics And The Creation Of An Internet Conscious User.
    Frishta Abdul Wali Princeton University Woodrow Wilson School of Public and International Affairs

    ABSTRACT

    The purpose of this policy analysis is to accomplish four goals: create a narrative around data retention in EU, give you a clear picture of the current public opinion regarding the existing policies, identify and analyze current policies, provide the Commission with guidelines in tackling the issue. Currently Data retention laws have taken a front seat in becoming a regulatory regime in political sphere as the digital market allows for free flow of information with little to no supervision of law enforcement. But with current imposed directives the EU has allowed for growing concerns within the public over personal data protection. National security and data privacy both impose upon one another and foster growing discourse. Identifying key issues that can highlight the legal issues of the policies, the lack of security in those policies and how the policies have continued to fail the constituents by allowing telecommunication agencies to exploit personal information will allow for the commission to understand the current narrative around data retention and what measures can be taken to alleviate the problem.

    Semi-Automatic Alignment Of Multilingual Parts Of Speech Tagsets

    Yashothara Shanmuagarasa1 and Uthayasanker Thayasivam2,1,2Department of Computer Science and Engineering, University of Moratuwa, Moratuwa, Sri Lanka.

    ABSTRACT

    We cast the problem of mapping a pair of Parts of Speech (POS) tagsets as a labeled tree mapping problem and present a general purpose semi-automatic POS tree alignment algorithm to solve the alignment. This algorithm can be used to align two POS tagsets of different languages or of same language. We evaluate its usefulness using POS tagsets of 2 languages, Tamil and Sinhala and provide the alignment between these languages. The proposed approach shows that manual effort in prior approaches is drastically reduced due to the proposed algorithm and also eliminates the need of creating new POS tagsets.

    Keywords

    Parts of Speech, POS tagset Mapping, POS tagset Alignment, Semi-Supervised Approach, BIS tagset, UOM tagset, Tamil NLP, Sinhala NLP

    Pervcompra-SE: A Pervasive Computing Reference Architecture From A Software Engineering Perspective.
    Osama M. Khaled, Hoda M. Hosny, and Mohamed Shalan Department of Computer Science and Engineering, The American University in Cairo, Cairo, Egypt

    ABSTRACT

    Pervasive Computing is a very challenging and complex domain that still lacks a comprehensive unified architecture. In this paper, we propose a reference architecture for pervasive computing that captures most, if not all, of the key challenges and provides a new architecture model that can be used in almost any business context. It provides conceptual views for the smart environment (SE), the smart object (SO), and the pervasive system (PS). We evaluated the model using a simulation prototype to predict its reliability at runtime.

    Keywords

    Pervasive computing, Ubiquitous computing, Context-aware services, Internet of Things, Software architecture, Reference Architecture, Model-Driven Architecture, Software design

    Optimized Kohenen - Self Organizing Map (K-SOM) Neural Network predictor for Software Defect Prediction.
    Kumudha.P Associate Professor Department of Computer Science and Engineering Coimbatore Institute of Technology, Coimbatore

    ABSTRACT

    The process of software testing is a vital task in software development environment and the defects detected at an appropriate time will result in saving time and the cost to be incurred for the development module. Software fault predictor algorithms are developed for identifying the software parts that are defect prone or that has got infected with bugs. Many early defect predictors have not considered misclassification cost of the defective and non defective modules. In this research work, software defect predictor model is developed by employing the versatile neural network algorithm, a neural model - Kohonen Self Organizing Feature Maps (K-SOM) that employs clustering approach to perform the defect diagnosing process. For all the datasets considered from the NASA Promise repository, K-SOM algorithm is applied individually and the results are observed. Further to this, the developed Hybrid GSA (Gravitational Search Algorithm) - CSSA (Charged System Search Algorithm) is employed in this work to tune the weight values of Kohonen Self Organizing Maps. The tuned weight values of K-SOM neuronal model employing proposed hybrid GSA - CSSA tend to achieve better results for both non-cost sensitive case and as well for cost-sensitive case in comparison with that of the earlier predictors available in the literature.

    Keywords

    Software defects - Neural Network-Kohenen Self Organizing Map (K-SOM) - Gravitational Search Algorithm (GSA) - Charged System Search Algorithm (CSSA)

    DSP Implementation Of A Fingerprint Algorithm.
    1Farah Dhib Tatar,2Mohsen Machhout 1Department of Electrical Engineering, National school of the studies of engineer of Tunis Tunis, Tunisia.2Faculty of Sciences of Monastir Monastir, Tunisia

    ABSTRACT

    Nowadays a fast and reliable identification of people is necessary for multiple applications such as access to the Internet, backup and use of personal data, access to environments and private places etc.Biometrics is an emerging field where technology improves our ability to identify a person. The advantage of biometric identification is that each individual has its own physical characteristics that cannot be changed, lost or stolen.Embedded systems are characterized by limited resources (calculation + energy). They are used everywhere in a multitude of applications (gaming stations, telecommunication, video applications, security, etc).This paper proposes a study of the performance of a DSP in image processing used for fingerprint biometric identification.

    Keywords

    Fingerprint; Biometrics; Images Processing; Embedded Systems; DSP; Computer Security; Computer Science.

    A HYBRID ACTIVE LEARNING AND PROGRESSIVE SAMPLING ALGORITHM
    Amr ElRafey and Janusz Wojtusiak, Department of Health Administration and Policy, George Mason University, Fairfax VA, USA
    ABSTRACT

    Sampling techniques for data mining applications can be broadly categorized into Random Sampling (RS), Active Learning (AL) and Progressive Sampling (PS). Progressive Sampling techniques grow an initial sample up to the point beyond which model accuracy no longer significantly improves. These methods have been shown to be computationally efficient. The sampling schedule to be used with progressive sampling techniques is still an ongoing issue of research due to the fact that available sampling schemes may either overshoot, resulting in a final sample which is larger than necessary, or they may grow the sample too slowly thus requiring many iterations of the algorithm before convergence is reached. We demonstrate how using Batch Mode Uncertainty Sampling from the domain of active learning, to progressively grow the sample, can significantly improve the performance of progressive sampling. Through a series of trials on both simulated and real data, we show that our proposed Progressive Batch Mode Uncertainty Sampling (PBMUS) algorithm converges with a smaller number of data points at higher accuracy and in some cases, less computational time.

    KEYWORDS

    Active Learning, Progressive Sampling, Uncertainty Sampling

    FPGA-IMPLEMENTATION OF WAVELET-BASED DENOISING TECHNIQUE TO REMOVE OCULAR ARTIFACT FROM SINGLE- CHANNEL EEG SIGNAL
    Chen Ronghua1 and Li Dongmei2, Zhang Milin2 1Department of Electronic Engineering, Tsinghua University, Beijing, China 2Department of Electronic Engineering, Tsinghua University, Beijing, China
    ABSTRACT

    This paper presents the real-time implementation on FPGA of the wavelet-based denoising technique to remove the ocular artifact from the signal-channel EEG signal. The advantage of this method over conventional methods is that there is no need for the recording of the electrooculogram (EOG) signal itself. This approach papers both for eye blinks and eye movements. Discrete Wavelet Transform (DWT) is selected end the hard-thresholding is applied to the wavelet coefficients using the Statistical Threshold (ST) estimated in interested bands. This real-time architecture presents two characteristics: 1) quantization of the filter coefficients and the elimination of the multiplier to reduce the hard cost, and 2) symmetrical extension of the signal boundary to full reconstruction while the data volume is invariable. Experimental results show that proposed architecture efficiently removes the ocular artifact from EEG signal..

    KEYWORDS

    Wavelet transform, EEG, ocular artefact, hard-thresholding, denoising

    EFFECT OF THE IMPROVED-SHAPED GATE IN HEMT TRANSISTORS PERFORMANCE
    Serveh Rahimi and Saeideh bahrebar,Department of Electrical Engineering, Razi University, Iran
    ABSTRACT

    In this paper the effect of the Gate shape on the operation of HEMTS is evaluated via simulations and Two transistors with different Gate Shapes are simulated. Simulations show that with the improved Gate-shaped (wedge-shaped) the performance of the transistor is improved. Therefore, when the Gate Voltage and Drain Voltage are increased, the curve Drain Current versus Drain voltage and Gate Voltage is also increased, means the transistors-conductance increases.

    KEYWORDS

    HEMT, GATE, PERFORMANCE, TRANSISTORS

Copyright © IPPR 2018