Specific Patches Decorrelation

Channel Feature on Pedestrian Detection

Xue-ming Ding and Dong-fei Ji

School of Optical Electrical and Computer Engineering, University of Shanghai for Science and Technology, China

Abstract: Typical Local Decorrelation Channel Feature (LDCF) for pedestrian detection generates filters derived from decorrelation for each entire positive sample, using Principle Component Analysis (PCA) method. Meanwhile, extensive pedestrian detection methods, which utilize statistic human shape to guide filters design, point out that the head-shoulder area is the most discriminative patches in typical classification stage. Inspired by above mentioned local decorrelation operation and discriminative areas that most classifiers indicate, in this paper we propose to integrate human shape priority into image patch decorrelation to generate novel filters. To be specific, we extract covariance from salient patches that contain discriminative features, instead of each entire positive sample. Furthermore, we also propose to share covariance matrix within grouping channels. Our method is efficient as it avoids extracting uninformative filters from redundant covariance of convergent patches, due to embedded prior human shape info. Experiments on INRIA and Caltech-USA public pedestrian dataset has been done to demonstrate effectiveness of our proposed methods. The result shows that our proposed method could decrease log-average miss rate with detection speed retained compared to LDCF and most non-deep methods.

Keywords: Specific patches partition, decorrelation, shared covariance, channel features, average human shape.

Received May 22, 2019; accepted May 28, 2020

Space-time Templates based Features

for Patient Activity Recognition

Muhammad Adeel Abbas¹, Fiza Murtaza^2,3, Muhammad Obaid Ullah¹, and Muhammad Haroon Yousaf²

1University of Engineering and Technology, Department of Electrical Engineering, Pakistan

2University of Engineering and Technology, Department of Computer Engineering, Pakistan

³Sino-Pak Center for Artificial Intelligence (SPCAI), PAF-IAST, Pakistan

Abstract: Human activity recognition has been the popular area of research among the computer vision researchers. The proposed work is focused on patient activity recognition in hospital room environment. We have investigated the optimum supportive features for the patient activity recognition problem. Exploiting the strength of space-time template approaches for activity analysis, Motion-Density Image (MDI) is proposed for patient’s activities when used supportively with Motion-History Image (MHI). The final feature vector is created by combining the description of MHI and MDI using Motion Orientation Histograms (MOH) and then applying Linear Discriminant Analysis (LDA) for dimensionality reduction. The LDA technique not only reduced the complexity cost required for classification but also played vital role to get best recognition results by increasing between-class separation and decreasing the with-in class variance. To validate the proposed approach, we recorded a video dataset containing 8 activities of patients performed in hospital room environment under indoor conditions. We have successfully validated the results of the proposed approach on our dataset by training the SVM classifier and achieved 97.9% average recognition accuracy.

Keywords: Human activity recognition, motion templates, patient monitoring, LDA.

Received September 3, 2019; accepted July 14, 2020

  Architecture Style Selection using Statistics of Quality Attributes to Reduce Production Costs

Hamidreza Hasannejad Marzooni¹, Homayun Motameni², and Ali Ebrahimnejad³

¹Department of Computer Engineering, Babol Branch, Islamic Azad University, Babol, Iran

²Department of Computer Engineering, Sari Branch, Islamic Azad University, Sari, Iran

³Department of Mathematics, Qaemshahr Branch, Islamic Azad University, Qaemshahr, Iran

Abstract: As time goes by and software systems grow in complexity and size, there is an increasing need for software architecture as an important tool in software design. Designing an appropriate architecture is necessary in producing a high-quality software, which also suits stakeholders. In order to design the desired high-quality software program, style-based architectures can be used. That is, with the selection of appropriate style architecture, we will get an ideal architecture for design. With the same attitude in this research, using a statistical computational algorithm, we have attempted to select the appropriate software architecture style to meet stakeholders’ requirements. In meeting Non-Functional Requirements (NFRs) of stakeholders, increase of one NFR does not increase the others necessarily, and they may be at odds with each other, thus the best quality for all cannot be achieved. In the designing stage of an ideal software, we must take into account the production and maintenance costs as well as a trade-off between stakeholders’ desired needs. The proposed algorithm structure involves a method using Gamma Probability Distribution Function (PDF). In a way that, a statistical estimate for each present style is created, and finally in the design of the software, the best style (based on the mentioned statistical estimate) is used for meeting the stakeholder’s needs. The method not only creates NFRs in the software program, but also gives importance to production and maintenance costs. This requires that the qualitative data of the problem be converted into quantitative data. It will be fully described in the introduction to the algorithm. In order to verify the validity of the proposed algorithm, the resulted architecture style ranking will be compared with the results of alternative methods namely Analytic Hierarchy Process (AHP) and A Lightweight Value-based Software Architecture Evaluation (LiVASAE). The results confirm the applicability of the proposed algorithm and moreover it has less time complexity with respect to other methods.

Keywords: Software architecture style, non-functional requirements, curve fitting, gamma method.

Received September 13, 2019; accepted June 18, 2020

Privacy Preserving Authenticated Key

Agreement based on Bilinear Pairing

for uHealthcare

Sunghyun Cho¹ and Hyunsung Kim²

¹College of Computing, Sungkyunkwan University, Korea

²School of Computer Science, Kyungil University, Korea

Abstract: With the growth of wireless communication technologies and sensor technologies, ubiquitous Healthcare (uHealthcare) based on Internet of Things (IoT) is becoming a big research focus from various researchers. However, security and privacy issues are top most important focuses to be solved for the success of uHealthcare services. This paper shows that Mahmood et al.’s authentication and prescription safety protocol is prone to denial of service attack and stolen-verifier attack. Furthermore, we propose a privacy preserving authenticated key agreement protocol for IoT based uHealthcare, which is based on hash function, symmetric key cryptosystem and bilinear pairing. The proposed protocol efficiently solves the security and privacy problems in Mahmood et al.’s protocol and also provides computational efficiency compared to the related protocols.

Keywords: Authenticated key agreement, authentication, internet of things, prescription safety, ubiquitous healthcare.

Received October 2, 2019; accepted October 25, 2020

https://doi.org/10.34028/18/4/4

Full text   

IOT-Pattern-As-a-Service Model for Delay

Sensitive IOT Integrated Applications

Murugan Sivaram

Research Center, Lebanese French University, Erbil, Kurdistan Region, Iraq

Abstract: At present, the Internet of Things (IoT) impacts heavily the daily lives of an individual in many domains, which ranges from wearable devices to industrial systems. Accordingly, these wide ranging IoT applications require application specific frameworks intended to carry out the operations in IoT applications. On other hand, IoT ecosystem evolves on integrating with other environments but the presence of heterogeneous devices in IoT integrated ecosystem groups the capacities in order to match the service requirements of users and to support wide users. Hence, a solution is required to synergize cooperation among the users in IoT integrated environment with great relevance. Along this line, the present work plans to adopt the IoT-Pattern-as-a-Service (IoT-PataaS) model to support Fifth Generation (5G) network environment, since the application is a delay-sensitive one and that should be controlled using high-end IoT devices. The proposed IoT-PataaS aims at provisioning IoT applications with reduced delay that leverages collaboration between the IoT objects in public and private clouds, which is present at the edge of 5G networks. The evaluation of IoT-PataaS model in 5G cellular network is carried out in terms of Narrowband IoT. The results claims that IoT-PataaS model obtains highly significant benefits in Narrowband IoT and LTE-A networks in terms of successfully delivered services in IoT platform.

Keywords: Internet of things, 5G network, delay-sensitive network, narrowband IOT, LTE-M and NB-IOT.

Received December 5, 2019; accepted January 4, 2021

A Robust Secure Self-Certified Concurrent Signature Scheme from Bilinear Pairings

Chien-Hua Tsai¹ and Pin-Chang Su²

¹Department of Accounting Information, Chihlee University of Technology, 220 Taiwan

²Department of Information Management, National Defense University, 112 Taiwan

Abstract: The idea of concurrent signature schemes is that two parties produce two respective ambiguous signatures that are concurrently bound to their corresponding signatories only while either of the party releases a keystone. The main construct is that both parties need to reach a consensus on the true fairness in mutually exchanging the signatures, and, moreover, the protocols assume that there is no collusion between a trusted third party and any of the parties. However, by collaborating over business interests with the participants as strategic partners, the trusted third party may obtain access to sensitive key data held in escrow, leading them to the collusion attack associated with malicious intentions. To circumvent the misbehavior among the participating individuals, an identity authentication process can be used prior to exchanging or having access to any confidential information. In this paper, we propose a self-certified concurrent signature from bilinear pairings as an alternative solution to strengthen the security level for solving the fair exchange problem. Apart from resisting to the collusion attack, the proposed scheme provides the advanced security properties to prevent from the message substitution, the identity forgery and impersonation, and other generic attacks in an increasingly insecure network environment.

Keywords: Bilinear pairings, concurrent signature, fair exchange, self-certified, trusted third party.

Received December 26, 2019; accepted February 21, 2021

Novel Turkish Sentiment Analysis

System using ConvNet

Saed Alqaraleh

Computer Engineering Department, Hasan Kalyoncu University, Turkey

Abstract: In this paper, an efficient model for the Turkish language sentiment analysis has been introduced. As Turkish is an agglutinative language, which requires spatial processing, an efficient pre-processing model was also implemented and integrated as a part of the developed system. In addition, the Deep Convolutional Neural Networks (ConvNet) have been integrated to build an efficient system.

Several experiments using the “Turkish movie reviews” dataset have been conducted, and it has been observed that the developed system has improved the sentiment analysis system that supports the Turkish language and significantly outperforms the existing state-of-the-art Turkish sentiment analysis systems.

Keywords: Sentiment analysis, opinion mining, text classification, turkish language, convolutional neural networks, natural language processing.

Received January 8, 2020; accepted February 21, 2021

Predicting Student Enrolments and Attrition Patterns in Higher Educational Institutions using Machine Learning

Samar Shilbayeh and Abdullah Abonamah

Business Analytics Department, Abu Dhabi School of Management, UAE

Abstract: In higher educational institutions, student enrollment management and increasing student retention are fundamental performance metrics to academic and financial sustainability. In many educational institutions, high student attrition rates are due to a variety of circumstances, including demographic and personal factors such as age, gender, academic background, financial abilities, and academic degree of choice. In this study, we will make use of machine learning approaches to develop prediction models that can predict student enrollment behavior and the students who have a high risk of dropping out. This can help higher education institutions develop proper intervention plans to reduce attrition rates and increase the probability of student academic success. In this study, real data is taken from Abu Dhabi School of Management (ADSM) in the UAE. This data is used in developing the student enrollment model and identifying the student’s characteristics who are willing to enroll in a specific program, in addition to that, this research managed to find out the characteristics of the students who are under the risk of dropout.

Keywords: Machine learning, predictive model, apriori algorithm, student retention, enrolment behaviour, association rule mining, boosting, ensemble method.

Received March 10, 2020; accepted July 19, 2020

Modeling of a PV Panel and Application of Maximum Power Point Tracking Command based on ANN

Mourad Talbi¹, Nawel Mensia², and Hatem Ezzaouia²

¹Laboratoire de Nanomatériaux et des Systèmes pour les Energies Renouvelables (LaNSER)

²Laboratoire Photovoltaique

Abstract: The Photovoltaic (PV) systems are renewable and environmentally friendly. However, they still have numerous disadvantages such as high investment cost and low efficiency. For getting highest efficiency from a PV system in different operation conditions, PV panels and arrays should be operated at Maximum Power Points (MPPs). At MPP, PV arrays produce the electric energy at maximum efficiency and minimum losses. Some algorithms are used in PV systems to provide maximum efficiency and minimum losses. In this paper, a new PV panel model is proposed employing Matlab/Simulink, and three commands of the MPP Tracking (MPPT), are applied in our proposed PV system. These MPPT commands are (P&O), the Incremental Conductance (IC) and the ANN based one. Then, a comparative study between these three commands is performed. The simulations results obtained from Matlab/Simulink software, are presented for these approaches under rapid variation of insolation and temperature conditions. Those results confirm the effectiveness of the ANN based command both in terms of efficiency and fast response time and this compared to the two other commands (P&O and IC). Also compared to P& O and IC, negligible oscillations around the MPP is the main advantage of this ANN based command.

Keywords: MPPT controller, photovoltaic panel, insolation, temperature, ANN.

                Received March 30, 2020; accepted September 7, 2020

Analysis of Alpha and Theta Band to Detect Driver

Drowsiness Using Electroencephalogram (EEG) 

Signals

Pradeep Kumar Sivakumar¹, Jerritta Selvaraj¹, Krishnakumar Ramaraj², and Arun Sahayadhas¹

¹Artificial Intelligence research Lab, Vels Institute of Science Technology and Advanced Studies, India

²Department of Electrical and Electronics Engineering, Vels Institute of Science Technology and Advanced Studies, India

Abstract: Driver drowsiness is recognized as a leading cause for crashes and road accidents in the present day. This paper presents an analysis of Alpha and Theta band for drowsinesss detection using Electroencephalogram (EEG) signals. The EEG signal of 21 channels is acquired from 10 subjects to detect drowsiness. The Alpha and Theta bands of raw EEG signal are filtered to remove noises and both linear and non-linear features were extracted. The feature Hurst and kurtosis shows the significant difference level (p<0.05) for most of the channels based on Analysis of Variance (ANOVA) test. So, they were used to classify the drowsy and alert states using Quadratic Discriminant Analysis (QDA), Linear Discriminant Analysis (LDA) and K- Nearest Neighbour (KNN) classifiers. In the case of Alpha band, the channels F8 and T6 achieved a maximum accuracy of 92.86% using Hurst and the channel T5 attained 100% accuracy for kurtosis. In the case of Theta band, Hurst achieved 100% accuracy for the channel F8 and Kurtosis obtained a maximum accuracy of 92.85% in the channels FP1, CZ and O1. A comparison between Alpha and Theta band for the various channels using KNN Classifier was done and the results indicate that the selected channels from Alpha and Theta bands can be used to detect drowsiness and alert the driver.

Keywords: Electroencephalogram, alpha band, theta band, drowsiness, KNN, ANOVA.

Received April 19, 2020; accepted December 1, 2020

A Real-Time Business Analysis Framework Using Virtual Data Warehouse

Partha Ghosh¹, Deep Sadhu², and Soumya Sen¹

¹A.K.Choudhury School of Information Technology, University of Calcutta, India

²Department of Data Science, Ernst and Young, India

Abstract: Data Warehouse (DW) is widely used in industries over decades to perform the analysis on data to expedite decision-making process. However, the traditional DW is slower in execution due to the huge time overhead of pre-processing stages of Extraction-Transformation-Loading (ETL). On the other hand, often the situations arise where the decision-making are required in real time. Data virtualization is one of the robust approaches over traditional data warehouse that avoids the costly steps of ETL processing. Virtual Data Warehouse (VDW) allows specific analysis for quick decision making even on the unprocessed data. Moreover, VDW could be used by the organizations that maintain DW to take some immediate business decisions for some abrupt changes. This research work performs business trend specific analysis based on VDW to generate business intelligence even in the catastrophic situations. Experimental results reveal, the proposed methodology based on VDW is around thousand times faster than traditional warehouses.

Keywords: Data warehouse, virtual data warehouse, data analytics, business intelligence, real-time analysis, z-score, query execution.

Received May 19, 2020; accepted December 15, 2020

A New Approach for Detecting Eosinophils in the

Gastrointestinal Tract and Diagnosing Eosinophilic

Colitis

Amal Alzu’bi¹, Hassan Najadat¹, Walaa Eyadat², Alia Al-Mohtaseb³, and Hussam Haddad⁴

¹Department of Computer Information Systems, Jordan University of Science and Technology, Jordan ²Department of Computer Science, Jordan University of Science and Technology, Jordan

³Department of Pathology and Microbiology, Jordan University of Science and Technology/ King Abdullah University Hospital, Jordan

⁴Department of Pathology and Microbiology, King Abdullah University Hospital, Jordan

Abstract: Eosinophilic Gastrointestinal Diseases (EGIDs) represent a rare group of disorders that can have various clinical presentations dependent on the involved segment within the gastrointestinal tract. Eosinophilic Colitis is considered as an under-diagnosed disease, which requires more attention and correct diagnosis. Our research aims to develop an image processing and machine learning approach that can be utilized by pathologists to diagnose patients with Eosinophilic Colitis in an easy and fast manner. The approach tends to enable pathologists to detect eosinophils in the microscopic sections of the gastrointestinal tract including; the esophagus and colon. We proposed an approach that relies on applying advanced image processing techniques on the digitally acquired images of microscopic biopsies to extract the primary features of the eosinophils and to estimate the count of the eosinophils in the given patient’s slide. These counts were used as inputs to machine learning algorithms including, Support Vector Machine (SVM) and Neural Networks in order to decide whether the patient has eosinophilic colitis disease or not. The accuracy of detecting Eosinophilic Colitis using SVM classifier is 85.71%, and in neural network is 93.8%.

Keywords: Eosinophilic colitis, eosinophils, eosinophilic gastroenteritis, image processing, digital images, neural network, SVM.

Received July 13, 2020; accepted December 10, 2020

Elitist Strategy of Genetic Algorithms for Writing

Tang Poetry

Wujian Yang¹, Wenyong Weng¹, Guanlin Chen¹, and Zihang Jiang²

¹Department of Computer and Computing Science, Zhejiang University City College, China

²College of Computer Science and Technology, Zhejiang University, China

Abstract: Automatic Chinese Tang poetry composition arouses researchers' attention these years and faces a lot of challenges. Most existing poetry generation systems can only generate poems without human interaction; thus, these poems cannot always express the human mind accurately. To improve this disadvantage, this paper proposes a modified elitist genetic algorithm to generate poetry with arbitrary interaction from the user, which means that the user can specify the poem’s emotion and input words or verses to be used in the poem. The modified algorithm comprises an improved elitist strategy to retain keywords or verses provided by the users, and a new concrete fitness function for more accurate and effective quality evaluation of poems. The Turing test and fitness function contrast experiments show that the proposed algorithm could generate poems using given keywords or verse and the poems generated by the algorithm receive higher ratings and recognition than the original poems written by a human. The experimental results demonstrate the effectiveness of the proposed algorithm and prove that this research can make practical and theoretical contributions.

Keywords: Elitist strategy, adaptive genetic algorithm, automatic generation, tang poetry, self-help writing.

Received October 24, 2019; accepted December 15, 2020

Abstract: ASAF is a multiple agent robotic system where Unmanned Aerial Vehicles (UAV) and Ground Vehicles (UGV) agents perform coordinated tasks. Our research group built this system based on Multiple Unmanned Autonomous Vehicle Experimental Testbed (MAUVET), an in-house platform that we have introduced as well. The challenge in the development of a mobile robotic system is that performance in real time deployment differs from the original plan. This case is clearer when planning the traversal path of an agent, where error happens because of mechanical and environmental factors. The aim of this paper is to investigate the agent traversal execution via system experimentation and computer simulation. The outcome of this investigation is understanding this behavior under different sets of circumstances and finding some optimization factors.

Keywords: Unmanned aerial vehicles, UAV controller, embedded systems, multilayered architecture.

Received June 18, 2019; accepted June 18, 2020