16th International Conference on Information Technology-New Generations (ITNG 2019)

Modular exponentiation represents a significant workload for public key cryptosystems. Examples include not only the classical RSA, DSA, and DH algorithms, but also the partially homomorphic Paillier encryption. As a result, efficient software implementations of modular exponentiation are an important target for optimization. This paper studies methods for using Intel’s forthcoming AVX512IFMA instructions in order to speed up modular (Montgomery) squaring, which dominates the cost of the exponentiation. We further show how a minor tweak in the architectural definition of AVX512IFMA has the potential to further speed up modular squaring.

Organizations and, to larger extent governments, have in recent years become more reliant than ever on the consistent operation of their information systems (ISs), which have become critical to their success and efficacy. Even though the growing dependence on ISs has created a pressing need for gathering information and easing its accessibility, security vulnerabilities in the systems have unfortunately spawned opportunities for hackers to compromise the system’s integrity and validity. This has put information systems confidentiality at high risk. As this paper shows, user authentication mechanisms, particularly passwords, determine confidentiality, which is an essential aspect of IS security. Passwords have been and are still a common method of securing ISs. A problem arises when users need to memorize and manage passwords that meet requirements, namely that users are more likely to use a single password across multiple accounts. User selection of passwords for use in ISs creates an impasse for the user. Furthermore, the systems progressive reliance on passwords for user authentication points to the necessity of investigating password creation policies in organizations. In spite of organizations recommendations on password choice to users, the majority do not choose to protect their passwords or even require choosing strong passwords in effect making hackers work easier. This study uses a survey questionnaire to evaluate user-generated password characteristics and present findings that prove user negligence with regards to password choice. Additionally, a literature review on password vulnerabilities in organizations is presented.

Cloud Computing is the new trend and this brings new issues and challenges in cyber security since a lot of the data from companies and users is available through the Internet. It is not only about the data but also about the applications that run in the Cloud that could be compromised affecting the service to thousands or millions of users, which may also have their local systems under siege through the exploitation of the security flaws in the Cloud. We propose here an algorithm that can detect unauthorized modifications to any of the files that are kept under custody. It is a collaborative systems where a set of nodes participate to gain also reputation points. This is a light algorithm that requires a time complexity of only O(n) times hashings in total. In our algorithm is implemented a technique to avoid the Hash Value Manipulation Attack that is one kind of Man-in-the-middle attack used to replace hash values. Any unauthorized modification of a file is detected and reported without the need of a third party auditor, which is another advantage.

This paper, compared three well-known technologies in cybersecurity: BlackRidge Transport Access Control, brain wave authentication technology, and SSL-VA. In addition to the determining the impact of these technologies on cyberspace, we compared them in the viewpoint of the CIA (Confidentiality, Integrity and Availability). Transport Access Control uses the first packet and defines a new real time to identify authorized users. The brain wave authentication technology is used in several ongoing projects that use grand-averaged technology and EEG to develop an authentication mechanism. SSL-VA uses advanced technology to decrypt packets between servers; it also benefits from input aggregation and output mirroring technology to decrypt packets. We also suggest local regulations and in-depth security control awareness when using these technologies and to improve their features.

Evidence collection of digital devices is the most important step of Forensic Investigation. We discuss four major evidence’s sources focusing on Cellphones, Global Positioning Systems (GPSs), Email and Network. Cellphone provides tremendous evidence, but it needs to be isolated to protect risk of being wiped up and overwritten. GPSs are valuable sources of information and can guide an investigator to identify “where the unit has been and where a user intended to go”. Network and Emails are the most powerful sources of evidence, but professional skills are needed to protect evidence from being overwritten and modified. In this paper using the CIA’s model, we prioritize these four sources of evidence. Network and email with 0.82 and 0.72 availability ratio received the highest availability score but with considering data integrity Cellphone and GPSs received the highest score and can be considered more reliable source of evidence during a Forensic Investigation.

The introduction of the processor instructions AES-NI and VPCLMULQDQ, that are designed for speeding up encryption, and their continual performance improvements through processor generations, has significantly reduced the costs of encryption overheads. More and more applications and platforms encrypt all of their data and traffic. As an example, we note the world wide proliferation of the use of AES-GCM, with performance dropping down to 0.64 cycles per byte (from ∼23 before the instructions), on the latest Intel processors. This is close to the theoretically achievable performance with the existing hardware support.Anticipating future applications and increasing demand for high performance encryption, Intel has recently announced that its future architecture (codename “Ice Lake”) will introduce new encryption instructions. These will be able to vectorize the AES-NI and VPCLMULQDQ instructions, on wide registers that are available on the AVX512 architectures. In this paper, we explain how these new instructions can be used effectively, and how properly using them can lead to the anticipated theoretical encryption throughput of around 0.16 cycles per byte. The included examples demonstrate AES encryption in various modes of operation, AEAD such as AES-GCM, and the emerging nonce misuse resistant variant AES-GCM-SIV.

Internet has become the main communication instrument between criminals. Expressions used by criminals are ciphered, by replacing language terms with regionalized and mutant expressions. There is a need to reveal, understand and formalize these obscure dialects to enable the automation of searches and the analysis of intentions. OntoCexp (Ontology of Criminal Expressions) aims at providing a common and extensible model for identifying usage of crime expressions in Internet. Its foundations come from an initial terminology and a semantic analysis of written communication between criminals (from Twitter) in Brazil (Portuguese language). 17 papers on ontologies, out of 63 articles of interest, have been selected and used as input to our proposal. The initial version of OntoCexp and its core elements are presented here; the complete ontology (OWL file) is available publicly to be used. We expect it to be useful for cyber-security researchers and criminal investigators who wish to formalize knowledge on criminal communication in their systems, methods, and techniques.

In an Internet of Things (IoT) environment some problems related to user privacy may occur because the exchange of information between devices occurs in a non-standard way. Brazilian data protection law – intended to protect personal data - must also be observed in applications implemented in the Internet of Things environment. This paper presents the IoTPC instrument, which is a tool for measuring privacy in IoT environments and is able to reflect users’ concerns with privacy. IoTPC consists of 17 items that understand users’ opinions on how some IoT devices collect, process, and make their personal information available in some specific IoT scenarios. The IoTPC tool was used in an inference model of the privacy negotiation mechanism for IoT systems. This model makes inferences based on IoTPC items and IoT scenarios using machine learning algorithms that have been trained and tested with IoTPC privacy preferences. The validation of the instrument was made by analyzing the result of a sample of 61 participants, considering the three first order dimensions (IoT requests, decision making and caution) through an exploratory factor analysis. The results of the learning process in the inference model had an accuracy of 79.20%, which indicates that IoTPC can be used in any privacy negotiation mechanism.

The convenient and cheap access to mobile phones and laptops have significantly increased the use of interactive applications over the past couple of years. However, this has posed various threats to legitimate users in terms of sensitive data disclosure, if their device gets lost, compromised or stolen. This study focuses on the forensic analysis of Windows AppStore applications with special focus on LinkedIn’s Desktop application; since it is one of the most downloaded applications from Windows AppStore. The paper first provides a systematic literature review of the existing digital forensic analysis techniques and highlights their weaknesses. A comprehensive novel methodology for manual forensic analysis of Windows App Store application on Windows 10 Operating System (OS) has also been proposed. For experimentation purpose, LinkedIn’s desktop application has been targeted. The research considers all kinds of scenarios such as logged in users, logged out users and intentional data deletion etc. It is finally concluded that from the viewpoint of application forensic analysis, the live, storage and registry analysis, all hold equal importance.

With the rise in digital crimes nowadays, digital investigators are required to recover and analyse data from various digital resources. Since the files are often stored in fragments owing to memory constraints, the information of the file system and metadata of the file is required to recover the file. However, in cases where the file system is destroyed intentionally or unintentionally, and the metadata is deleted as well, the recovery of the digital evidence is done by a special method known as carving. In file carving, files are recovered solely based on the information about the structure and content of the individual file rather than matching the system’s information of the file. The process of file carving in digital forensics first requires classifying and then arranging the blocks of data that are typically stored as a sequence of bytes in memory. But carving is only possible when the file is not damaged or corrupted otherwise carving is not possible. The aim of this research is to analyse various Windows OS’s file carving techniques used in Digital Forensics particularly for their strengths and weaknesses. This analysis leads to the need of explicitly designed file carvers for different types of files. A novel technique for carving Microsoft’s Word files (a compound format file which is least researched upon) has also been proposed in the document.

There is a large and fast growing demand for cybersecurity professionals who are well prepared and qualified to perform the challenging work of defending the cyber space. This paper explores and discusses the significant value and benchmark role of the Certified Information Systems Security Professional (CISSP) certification in the competency development for cybersecurity workforce by analyzing the CISSP certification requirements and objectives and mapping them to the US cybersecurity industry model of competencies and the US national cybersecurity workforce framework (NCWF). This paper also discusses the value and implications of the CISSP certification on cybersecurity education and training curriculum.

There has been a large and increasing demand for professionals and leaders in the cybersecurity field. Graduate schools in America are facing the challenge and opportunity to meet the demand by discovering talent and producing qualified workforce for the cybersecurity industry. This research paper proposes a model of discovering student interest and talent in graduate cybersecurity education through career guidance and curriculum development. This study is to illustrate the model with sample graduate student admissions and graduate cybersecurity program data at a private university in the northeast region of the United States.

In September of 2016, a new threat appeared on the internet that launched crippling Denial of Service Attacks against several high-profile targets. The Mirai Botnet, as it was called, took advantage of the weak security measures on Internet of Things (IoT) devices and used them to launch these DDoS attacks. The severity of these attacks awakened the technology industry to the lack of security for IoT devices. The use of IoT devices is expanding rapidly, creating an increasingly large attack surface for threats like Mirai. Multiple technology organizations have since been working to develop standards to push manufacturers to emphasize security on their devices and implement new technologies to improve security for consumers and enterprises. The renewed interest in security and the development of security standards may not be enough, as the market for IoT devices often incentivizes ease of use and practicality over security. The goal of this document is two-fold. First, I will explore the workings of one of the most disruptive pieces of malware in recent history, the Mirai Botnet. Second, I will explore methods to secure IoT devices from being compromised by malware such as the Mirai Botnet.

Software security is an important and a prevalent element in today’s society. The System Development Life-Cycle (SDLC) process that is currently used for most of software development does not address any security components until after the software is developed. The Secure Software Development Life-Cycle (SSDLC) is similar to the SDLC but includes security components into the phases. There have been many models proposed that are primary modified from preexisting SSDLC models. A study was conducted to survey different SSDLC models and their effectiveness. The authors first identified four popular SSDLC models in the IT industry, and then analyzed their common characteristics to derived four sets of Criteria for comparison. These criteria are: Focus Areas of Application, Implementation of model, Security Implementations and Enhancements, and Security training and Staff. The comparison results demonstrate that the Rastogi and Jones model is an effective one for being used for many IT projects. However, it is worthy to mentioned that one specific model cannot be used for all types of IT projects because IT projects are different.

Context: Money Laundering (ML) is a global crime that has a close relation with other crimes, such as: illegal drug trading, terrorism or arms trafficking. Criminals in today’s technology-driven society use every means available at their disposal to launder the profit made from their illegal activities. In response, international anti-money laundering (AML) efforts are made with AML systems. Objective: Identify and systematize the approaches, techniques and algorithms used in Computer Science (CS) to fight ML, besides identifying the trends in the field. Method: A systematic literature mapping was conducted to analyze the scientific research in the field. Results: The main approaches were identified, supervised classifiers and clusters, along with the trend of papers published over the years. China was the country with the highest number of published papers. Conclusion: The most relevant studies in such research line adopt data mining and machine learning techniques using clusters and classifiers. The state of the art was mapped, making it clear that it is an area of interest for researchers around the world with growth potential. We believe that this work is relevant to the academy, governments and the community at large, presenting them with trends in the detection of money laundering.

This paper focuses on the viability of using smart contract controlled electronic locks to give access to people by the lock’s owner in locations like hotels, coworking spaces or even their homes. The goal is that the guest rents a property using the lock and receives a temporary digital key according to his reservation which he uses to unlock the property’s door using an app on his smartphone. The lock was designed using a Raspberry Pi 3 B+ controlling a 12v solenoid lock through a python script and running the chirp.io audio communication protocol for both microcomputer and smartphone. The smart contracts were developed using Solidity from the Ethereum platform. The paper will cover the designed architecture and implementation details of the project.

The concept of biometric authentication is popular in the research industry. Biometric authentication refers to the measurement and statistical analysis of a human’s biological and behavioral features. Biometric technology is mainly used for authentication and identifying individuals based on their biological traits. Regarding biometric applications, security is the key issue that has a lot of remaining challenges. To succeed in this domain, this paper gives a background on the fingerprint matching algorithm steps. Moreover, the paper presents a brief overview of different attacks and threats that affect the privacy and security of the biometric system. Then we discuss the common schemes that have been used to secure biometric systems. Finally, findings and direction for further research about biometric system security are explored.

Connecting and controlling the access to the Internet of Things (IoT) are crucial since it converges and evolves multiple technologies. Traditional embedded systems, wireless sensor networks, control systems, software-defined radio technologies, and smart systems contribute to the Internet of Things. The deep learning, data analytics, and consumer applications have an essential role to IoT. The challenges are to store, process, and develop a meaningful form so that it is useful to the business, government, and customers. The paper discusses computing the data in a cloud; current challenges to store, retrieve, and process; security requirements; and possible solutions. We further provide the trust framework for the user and access control algorithms for data processing in the cloud environment.

As we enter the quantum computing era, security becomes of at most importance. With the release of D-Wave One in 2011 and most recently the 2000Q, with 2,000 qubits, and with NASA and Google using D-wave Systems quantum computers, a thorough examination of quantum computer security is needed. Quantum computers underlying hardware is not compatible with classical boolean and binary-based computer systems and software. Assemblers and compliers translate modern programming languages and problems into quantum-annealing methods compatible with quantum computers. This paper presents a vulnerability assessment utilizing static source code analysis on Qmasm Python tool. More specifically, we use flow-sensitive, inter-procedural and context-sensitive data flow analysis to uncover vulnerable points in the program. We demonstrate the Qmasm security flaws that can leave D-Wave 2X system vulnerable to severe threats.

Currently the Project Management Institute (PMI) estimates that approximately 25% of the world’s Gross Domestic Product (GDP) is spent on projects of various kinds and that about 16.5 million professionals are directly involved in project management worldwide. This volume of projects and changes in the world scenario, increasingly competitive, generate the need for faster results, with higher quality, lower costs and shorter deadlines. Among the main techniques for analyzing cost, time and scope performance, the Earned Value Management (EVM) technique is considered to be the most reliable. Several formulas derived from EVM’s measurements are available and have been studied over the past 15 years. However, EVM has a significant limitation regarding quality in its method. The technique is effective in providing cost and schedule related information but is still weak in taking the quality factor into account. The main objective of this work is to contribute to studies that seek to add the quality component into EVM and comparing performance between them. This paper presents the results of a systematic review, providing a comprehensive summary of the main problems with the use of the EVM technique and the possible solutions found to improve its capacity to predict the impact of quality (possible bugs or nonconformities) in the course of a project’s life cycle.

During the 1st Semester of 2018, at the Brazilian Aeronautics Institute of Technology (Instituto Tecnologico de Aeronautica – ITA), a successful Collaborative Interdisciplinary Problem-Based Learning (Co-IPBL) experience took place. At that time, more than 20 undergrad and graduate students from 3 different courses within just 17 academic weeks had the opportunity of conceptualizing, modeling, developing, and testing a Computer System based upon Big Data, Blockchain Hyperledger, Micro-services, and other emerging technologies for government and private organizations. This Co-IPBL was performed with the participation of a medical technical team from the Hospital of Clinics at the Faculty of Medicine of the University of Sao Paulo focusing in Urgency and Emergency Care. This research work was stored in a Google site and implemented as a Proof of Concept (PoC), by using emerging technologies such as the Blockchain Hyperledger. It represents one example of how to address the old problem of teaching, learning, designing, and implementing complex intelligent systems to solve health care problems, by collaboratively working with medical technical teams facing real problems of Urgency and Emergency Health Care.

During the 2nd Semester of 2018, at the Brazilian Aeronautics Institute of Technology (Instituto Tecnologico de Aeronautica – ITA), a successful Collaborative Interdisciplinary Problem-Based Learning (Co-IPBL) experience took place. At that time, more than 20 undergrad and graduate students from 3 different courses, within just 17 academic weeks, had the opportunity of conceptualizing, modeling, developing, and testing a Computer System involving multiple actors (Patients, Doctors, Hospitals, and Suppliers) for real-time decision making in the rehabilitation with Exoskeletons of patients suffering from Lower Limb Impairment after motorcycle accidents. Differently from other existing products from universities, research centers, governmental agencies, and other public and/or private companies, this product was developed, using the best practices of the Agile Scrum Method, along with emerging Information Technologies (ITs) such as Blockchain Hyperledger, Internet of Things (IoT), among others. This Co-IPBL was performed with the participation of a rehabilitation medical team from the Hospital of Clinics at the Faculty of Medicine of the University of Sao Paulo (HC-FMUSP). The experience described in this paper illustrates a way of dealing with the multiple challenges involved in teaching, learning, designing, and implementing complex intelligent systems to address health care issues with collaborative work involving multidisciplinary teams facing real-life problems such as exoskeletons applied to clinical recover of Patients.

In this paper, we compare two formal model-based frameworks supporting both system modeling and simulation which are based on models of computation: Ptolemy II and ForSyDe. The paper shows the main benefits and drawbacks of each compared framework, and also brings two different systems modeled using the synchronous reactive and synchronous dataflow models of computation.

One approach widely adopted for software testing rely on mutation to use as their comparison criteria. However, in mutation testing, the number of useless mutants generated during the mutation process can reach up to 90% of the total number of mutants created, making the process costly and time-consuming. The concept of subsumption has been empirically used in mutation testing to establish fault hierarchies in which tests that detect certain failures guarantee the detection of others, drastically reducing the generation of these useless mutants during mutation analysis. The tools used for mutation testing do not yet present automated solutions for checking for dominance and, consequently, selection of mutants to be considered in the tests. Hence, in this paper we present a model of automatic identification of subsumption in mutation testing based on evolutionary algorithms, demonstrating evidence of the efficiency of the use of artificial computational approaches for automatic generation of graphs of dynamic dominances in complex problems. We consider in our approach the dynamic subsumption method and a visualization model from the generation of graphs.We show that our approach is feasible and useful.

Currently, virtual and real social networks overlap through a wide access to Internet-connected devices, which use software and websites that promote interactions among individuals, groups and networks of dynamic relationships. However, it is important to investigate the acceptance and appropriateness of these alternatives in relation to different users’ profiles. This research aims to investigate the perception of a users’ group facing the possible adoption of the Life360 application seeking improvement of family safety. It is a qualitative study conducted by means of questionnaires and interviews with people from the city of Itajubá – MG. The perceptions collected were analyzed from the perspective of the Diffusion of Innovation Theory and the progressive model of adoption. As a result, a low-cost model that allows evaluating critical points in the process of adoption was obtained. It is concluded that Life360 is more suitable for people with higher income and education, given the features found in the progressive model of adoption.

Context: The literature provides evidence of challenges and difficulties related to the migration of legacy software systems to a microservice based architecture. The idea of microservices stipulates that the software be organized as a suite of small, modular, and independently deployed services, in which each service runs on its own process and communicates through well-defined, lightweight mechanisms to serve a business goal. However, the literature lacks step-by-step guidelines telling practitioners how to accomplish the migration from an existing, monolithic structure to a microservice based architecture. Goal: Discuss lessons learned from the migration of legacy software systems to microservices-based architecture. Method: We conducted two studies (a pilot and a case study) aiming at characterizing the relevants steps of such guidelines. Results: We report the steps and challenges observed during the migration reported in this study. Conclusion: We identify at least three main phases that drive the migration process.

This article presents an approach for performance verification of requirement scenarios in Service-Oriented Architecture (SOA) models. The SOA models are represented by Interorganizational WorkFlow nets that are not necessarily deadlock-free. The requirement model considered in this article defines functionalities that are of interest of all parties involved in the process, for this reason is considered as a public model. An architectural model is considered as a private model since it is composed of a set of private processes that interact through asynchronous communication mechanisms. For the scenarios of the architectural and requirement models that are equivalent in terms of behaviour (or bisimulation) symbolic dates associated with the activities of the processes can be produced from the calculation of the Linear Logic sequents. Symbolic dates can be used to verify if an architectural model scenario that simulates the behavior of a requirement model scenario is also equivalent in terms of performance. To illustrate the proposed approach an example was considered and from this example it was possible to observe that the approach can be useful to indicate if a architectural model in the context of SOA satisfy the performance of business needs defined by a requirement model.

Context: Opinions given by hotel clients in tourism social networks, the ones which can be a great source of knowledge extraction in the Big Data context, including the sustainable aspects of the hotels clients opinions. Objective: Evaluate performance and quality of deep learning neural networks, especially the Target-Connection LSTM (TC-LSTM) and Attention-based LSTM (AT-LSTM) algorithms, aiming to mine and classify the opinions posted on the TripAdvisor and Booking social networks, by considering sustainability aspects. Method: A controlled experiment to compare the efficiency and efficacy of the classifiers was carried out. Results: The AT-LSTM algorithm presented the best results, especially in terms of accuracy, precision, f-measure, average training time and average classification time. The first with 74,58%, the second with 95,54%, the third with 85,37%, then fourth with 7,3 s and the last one with 1,12 s. Conclusion: The AT-LSTM algorithm was expressly more effective than TC-LSTM, making it an option to be considered for mining opinions based on specific aspects of tourism and peculiar market niches.

Solar flares are huge releases of energy from Sun. They are categorized in five levels of potential damage to Earth (A, B, C, M, and X) and may produce strong impacts to communication systems, threatening human activities dependent on satellites and GPS. Therefore, predicting it in advance may reduce their negative impacts. However, solar flare forecasting has significant challenges: (a) the data is highly imbalanced, (b) adjacent classes are sometimes indiscernible, (c) current approaches perform binary forecasting (aggregating solar flare classes), instead of multi-class, as actually required. So, we aim to tackle these open issues by proposing a new Ensemble method called ECID(Ensemble of classifiers for imbalanced datasets). For each solar flare class, ECID employs a stratified random sampling for training one-class base inducers, strengthen their sensitivity to a given class. Using a modified bootstrap approach, the aggregator method combines the inducers results, enabling multi-class forecasting, which can also be multi-label in case of indiscernible classes. The results showed that ECID is well-suited for forecasting solar flares, achieving a maximum mean of True Positive Rate (TPR) of 96% and a Precision of 87%, in a time horizon of one day.

There is a great challenge in carrying out multivariate process capability analysis and fault diagnostics on a high dimensional non-normal process, with multiple correlated quality characteristics, in a timely manner. This paper proposes a hybrid capable of performing process capability analysis and fault diagnostics on multivariate non-normal processes. The proposed hybrid first utilizes the Geometric Distance (GD) approach, to reduce dimensionality of the correlated data into fewer number of independent GD variables which can be assessed using univariate process capability indices (PCIs). This is followed by fitting Burr XII distribution to independent GD variables. The independent fitted distributions are used to estimate both yield and multivariate process capability in a time efficient way. Finally, machine learning approach, is deployed to carry out the task of fault diagnostic by identifying and ranking the correlated quality characteristics responsible for the poor performance of individual GD variables. The efficacy of the proposed hybrid is assessed through a real manufacturing example and four simulated scenarios. The results show that the proposed hybrid is robust in estimating both yield and multivariate process capability carrying out fault diagnostics beyond GD variables, and identifying the original characteristic responsible for poor performance.

Nowadays, effective use of technology in the competitive world is the most important factor in business success. Since the stakeholders and especially customers are the most important factor in the profitability of organizations, the use of new methods in determining their satisfaction is of particular importance. In this paper, business intelligence for determining customer image satisfaction using image mining technology was performed. First, 400 images of customers’ faces were selected from the database. Investigating the facial expressions and the use of experts suggests that 4 states of 7 faces are sufficient to determine satisfaction. In the next step, the results of the study were extracted with using three algorithms and were analyzed in the following with three Classification Regression Tree (CRT) algorithms, Neural Network and Decision Tree. In the end between image mining algorithms the Bayesian Method and among data mining algorithms the CRT algorithm were detected with the least error for satisfaction detection.

With a greater ability for Big Data to obtain and compute such data sets, came a loss in the ability to ensure accurate data collection using traditional methods. This study aims to develop a model for data integrity, accurately detecting and compensating for errors and data drift. We will be working with a real-time water monitoring device that measures data in regular intervals. The data is managed in a decentralized manner by a Cloud-based Database Management System. We begin by exploring various drift compensation techniques. Afterwards, several algorithms implemented in similar studies are investigated. Upon choosing the optimal algorithm with any proper modifications, it is then implemented into an environmental sensing web application. In addition, new data visualization techniques are proposed to promote awareness for when data begins to drift.

We propose an approach to quantify the extent to which vertices in a complex network are clusterable on the basis of the similarity of the values with respect to two or more node-level metrics. We use the Hopkins Statistic to assess the clusterability and consider the centrality metrics as the node-level metrics. Our approach is to construct a logical topology of the vertices in the complex network using the normalized values of the centrality metrics as coordinates and determine the Hopkins Statistic for such a logical topology of the vertices. Our hypothesis is that if two or more vertices in a complex network have similar values for the centrality metrics, then the vertices should be clusterable to one or more clusters due to their proximity to each other in the normalized centrality-based coordinate system. The value for the Hopkins Statistic measure (ranges from 0 to 1) for such a logical topology of the vertices should be high, and vice-versa. We evaluate the Hopkins Statistic for 47 real-world complex networks (of diverse degree distributions) with respect to the neighborhood-based and shortest path-based centrality metrics. We observe the neighborhood centrality-based logical topologies of the vertices to incur relatively larger Hopkins Statistic values (i.e., exhibit higher similarity-based clusterability) for a majority of the networks.

Today the Web is pervasive in everyday life. Daily activities such as shopping and banking are now available from almost everywhere, which makes modern life more convenient. However, not everyone may benefit from this convenience. Low web literacy still prevents many users to take full advantage from online services. A group that usually presents issues related to web access is the elderly. As people grow older, motor control, visual acuity and cognition decreases, which makes aging users struggle to perform tasks in web applications. Therefore, it is important to detect struggling web users in order to support them, for instance, by providing friendly user interfaces. In order to tackle this problem, we propose an approach that is able to identify usage patterns commonly found among the elderly. Our approach allows the identification of struggling users while they browse web applications. Thus, by using our approach, developers may code adaptations to support these users. An experiment performed with real data from an educational web site shows that our approach is effective to identify struggling users in web applications.

The agriculture sector is facing major challenges to enhance production in a situation of dwindling natural resources. The growing demand for agricultural products, however, also offers opportunities for producers to sustain, improve productivity and reduce waste in the supply chain. Recent advancements in information and communication technologies (ICT) show promise in addressing these challenges. This paper proposes food security architecture for agricultural supply chain efficiency using ICT tools such as Internet of Things and Big Data analytics. To avoid loss in agriculture, a food security architecture is developed through Smart Agribusiness Supply Chain Management System is developed in this paper. This can result in uplifting the livelihoods of the rural poor through the enhancement of agribusiness.

Creating your own musical pieces is one of the most attractive ways to enjoy music. However, many musically untrained people lack the basic musical skills to do so. In this paper, we seek to explore how machine learning algorithms can enable musically untrained users to create their own music.To achieve this, we propose a Neural Hidden Markov Model (NHMM). It is a hybrid of a Hidden Markov model (HMM) and Convolution neural network (CNN) with a Long Short-Term Memory (LSTM) neural network. This model takes users’ original musical ideas in an easy intuitive way and automatically modifies the input to generate musically appropriate melodies as output. We further extend the model to allow users to specify the magnitude of revision, duration of music segment to be revised, choice of music genres, popularity of songs, and co-creation of songs in social settings. These extensions enhance user understanding of music theory, enrich their experience of self-learning, and enable social aspects of music creation. The model is trained using MIDI files of existing songs. We also conduct experiments on melody generation.We also hope to design a mobile application with an intuitive, interactive, and graphical user interface, which is suitable for the elderly and young children. Different from most existing literature focusing on computer music composition itself, our research and application aim at using computers to aid human composition and enriching the music education of musically untrained people.

The demand and use of automatic test generators for exams and competitions have both been increasing. But the handiness offered by these generators brings a difficulty in the process of creating databanks of questions: from a group of questions their different versions are still made for each question separately. Of course, this manual task turns out to be lengthy and costly. This problem proved to be effectively solved by generators of parametric questions. Our present work introduces an innovative and accessible approach that consists of an open platform to generate and correct parametric questions. To the best of our knowledge the platform presented here is the first one totally devoted to exams in hard copy that also includes automatic correction. Teachers and professors of any institution that registers in our platform can share question databanks among them.

This research discusses the enhanced three-tiered structure used in the NEST project model which has been designed to recruit and retain interns and scholars to become qualified STEM teachers. Based on existing program data and external program evaluations through surveys and interviews from the existing NSF Noyce Scholarship Phase I program, several key findings were used to enrich the design of the NEST project. Several improvements were implemented from the current model.

In one of its roles, Robotic seeks to stimulate the knowledge and learning in science areas to teenagers, young people and adults, trying to bring them closer to mathematical and physical problems of the real world. To do that, they need a very robust platform that allows build, play and validate real experiments. The target of this paper is to present an architecture that allows a Robotic Player built using a Bioloid ROBOTIS Premium platform to play soccer in the category of Kid-Size RoboCup Humanoid League. This architecture is composed by an Electronic Project, Vision System and Robotic Player. In the Electronic Project, a shield named Raspi2Dynamixel was developed to connect the Dynamixel AX-12A servo motor from ROBOTIS with Raspberry Pi 3B. Also, the AX12-JavA library was created to make the communication between them. In the Vision System, a webcam was connected to the USB port in Raspberry Pi 3B. With the proposed architecture, it is possible to expand the applications using the AX-12A servo motors, not only for humanoid soccer player but also for other applications, especially in the Educational Robotic field. In order to prove that the proposed architecture works well, the movements of the Robotic Player were shown in the results. In addition, it is important to highlight that has not found in the existing literature any other shield that can perform the communication between Raspberry Pi 3B and AX-12A servo motors using Java language.

In multicore or heterogeneous platforms, the last level cache (LLC) is usually shared by multiple resources and its performance has a significant impact to overall performance. Effective utilization of the LLC helps in reducing the effect of costly miss penalty. As the number of cores and threads per core increases, the number of sets utilized at certain point of time is limited, which is leading to cache thrashing. In this paper, we propose an efficient cache data placement of last level cache which reduces the effect of cache thrashing. It is observed that the decrease in cache performance of multithreaded applications is due to inefficient usage of cache sets. The initial performance analysis is done by running 6 benchmarks from PARSEC 2.1 and SPLASH2 benchmark suites. The miss traces of L1 caches are obtained and a trace driven L2 cache simulator. Miss rate reduction is observed from the proposed data placement method. Also, we see the effective usage of sets with the proposed cache thrashing reduction scheme.

Missing data may be one of the biggest problems hindering modern research science. It occurs frequently, for various reasons, and slows down crucial data analytics required to answer important questions related to global issues like climate change and water management. The modern answer to this problem of missing data is data imputation. Specifically, data imputation with advanced machine learning techniques. Unfortunately, an approach with demonstrable success for accurate imputation, Fuzzy K-Means Clustering, is famously slow compared to other algorithms. This paper aims to remedy this foible of such a promising method by proposing a Robust and Sparse Fuzzy K-Means algorithm that operates on multiple GPUs. We demonstrate the effectiveness of our implementation with multiple experiments, clustering real environmental sensor data. These experiments show that the our improved multi-GPU implementation is significantly faster than sequential implementations with 185 times speedup over 8 GPUs. Experiments also indicated greater than 300x increase in throughput with 8 GPUs and 95% efficiency with two GPUs compared to one.

While quantum computing has shown great promise in the field of computer science, a lack of actual practical quantum hardware means that mainstream research must rely on simulations. As such, a wide number of quantum computing simulation libraries have been developed, each with their own strengths and weaknesses. A good simulator must not just be accurate, but fast as well. This is especially relevant for quantum systems since the problem size growth for quantum systems is super-exponential. For this paper, we introduce a quantum computing simulation system that takes advantage of multiple gpus to achieve up to 400 times faster simulation time. We discuss the implementation details and provide analysis of its performance. We also demonstrate how the real-world phenomenon of quantum gate incoherence can be accurately simulated by varying the floating point precision and demonstrate it by using a precision of 9 bits, which we evaluate using Randomized Benchmarking.

Capsule Networks are an emerging extension of the traditional multilayer perceptron model, providing classification and estimated pose parameters through extra computational data models such as vectors and matrices. With extra data comes extra processing and meticulous data manipulation and this paper presents a scalable GPU optimization for the training and evaluation of such networks. A novel data abstraction maps the individual values of this tensor to one-dimensional arrays to 2D grids of multi-dimensional elements, enabling the removal of redundant data movement and other structural overhead found in common machine learning libraries. Assuming a single loss function, adapted for vector-based forward and backward propagation, this optimization involves combining successive operations found in processing into kernel calls, albeit at the expense of higher memory resource utilization. Despite that, this GPU based approach reached 33 times speed up in forward propagation and 130 times speed up in back propagation against optimal single threaded CPU calculations.

Image analysis is an important area in many fields of research, such as sensor networks, where webcams have become an increasingly popular addition. Sensor network webcams gather images frequently and, as such, a need for processing large image streams has occurred. In this paper, we test a variety of OpenCV functions on image streams from sensor networks on CPU, GPU and multiple GPU to find the most efficient way of processing image series. OpenCV is a popular computer vision and image processing library. Although OpenCV supports GPU functionality, the library’s multiple GPUs functions are lacking. Two image processing algorithm for snow and cloud detection were developed using OpenCV and applied to an image series. The execution times, speedup, and throughput of the different implementations are compared and discussed. From this, The best execution method for various situations is determined.

This work presents a parallel implementation of the Two-Dimensional Detrended Fluctuation Analysis (2D-DFA) algorithms using GPGPU-CUDA through the PyCUDA Wrapper. Both the monofractal and multifractal versions of 2D-DFA present poor performance in their sequential implementations because of high computational complexity, rendering ineffective the use for real-time applications and those classified as Big Data. However, these algorithms present the requirements needed to be parallelized using GPUs. Mainly because the same computation is performed in different sub-matrices of the data. Therefore, the concepts of data parallelism and independence, necessary for computation in GPUs were attending. The parallelization strategy was process simultaneously several sub-matrices, and not parallelizes the internal procedure in each sub-matrix. Matrices with different dimensions and Hurst exponents were generated for the validation of the method implemented. The results showed that both algorithms considerably improved their processing times, confirming the hypothesis that the parallelization using GPUs is a good strategy to improve the performance of the studied algorithms. The speed up to monofractal version was better than multifractal version.

Many complex real world systems can be represented as correlated high dimensional vectors (up to 20,501 in this paper). While univariate analysis is simpler, it does not account for correlations between variables. This omission often misleads researchers by producing results based on unrealistic assumptions. As the generation of large correlated data sets is time consuming and resource heavy, we propose a graphical processing unit (GPU) accelerated version of the established NORmal To Anything (NORTA) algorithm. NORTA involves many independent and parallelizeable operations—sparking our interest to deploy a Compute Unified Device Architecture (CUDA) implementation for use on Nvidia GPUs. NORTA begins by simulating independent standard normal vectors and transforms them into correlated vectors with arbitrary marginal distributions (heterogenous random variables). In our benchmark studies using a Tesla Nvidia card, the speedup obtained over a sequential NORTA coded in R (R-NORTA) peaks at 19.6× for 2000 simulated random vectors with dimension 5000. Moreover, the speedup obtained for GPU-NORTA over a commonly used R package for multivariate simulation (the copula package) was 2093× for 2000 simulated random vectors with dimension 20,501. Our study serves as a preliminary proof of concept with opportunities for further optimization, implementation, and additional features.

CRISPR is a mechanism used by archaea and bacteria to prevent virus attacks. The CRISPR-Cas9 technique was inspired by this natural genome editing system in bacteria where the latter captures snippets of DNA from invading viruses to create DNA segments known as CRISPR arrays. CRISPR-Cas9 aims to detect RNA virus pattern locations in DNA sequences and then use them to deactivate attacking viruses. Many CRISPR-Cas9 tools and algorithms were implemented for genome editing to locate viruses. However, some of these algorithms have poor performance which hinders their applicability on large DNA sequences. In this paper, we use a well-known sequence alignment algorithm, Smith Waterman, to calculate the similarity between a pattern query and long DNA sequences. Our approach has four versions which are sequential, parallel CPU, parallel GPU and hybrid CPU-GPU implementations. Our experiments show that the hybrid implementation gets the best results with performance gain reaching 25 times faster than the sequential implementation and 6 times faster than the parallel CPU implementation.

In this study, the statistical models project (SMp) proposes a six-parameter probabilistic function of two steps that can generate probabilistic functions (PFs) for a random continuous/discrete variable X, SMp(x); showed the actual widely used binomial (BD), Poisson and normal distributions, their deficiencies and limitations and the little or non-probabilistic importance of the former; and developed the first version of the computer tool associated with the proposed function, which is a MATLAB application of two modules. One of these modules models stochastic data (x;y), such as those that follow the identical or similar behaviors of the normal, Poisson, BD and others, and those associated with stochastic processes/effects (SP/Es), such as, normal tissue complication, percentage depth dose, and pharmacokinetic. The second module calculates the probabilities for a random continuous variable X using SMp(x).The proposed function can be used in the role of some probabilistic distributions (PDs) or probability density functions (PDFs), and overcome their deficiencies and limitations. SMp(x) generates three SMp types of the SP/Es if variable x is replaced for y. For its probabilistic conditions, at least one SMp(x) parameter depends on the others. One of the main objectives of this study is showing that the BD is actually a mathematic exercise and creation, and the advantages of the SMp in its role SMp-normal over the Gaussian distribution.

Identification of genetic components that are either shared by or differentiate multiple disorders is one of the most crucial issues for understanding the underlying biology of disorders and improving diagnosis and treatment based on genetic features. Although one of the most promising types of studies for this purpose is the genome-wide association study (GWAS) across disorders, only a modest fraction of the total heritability can be explained by loci identified through GWASs, partly due to a lack of sufficient statistical power. To search for relevant components more effectively, we developed an alternative efficient multi-subgroup gene screening method using multidimensional semiparametric hierarchical mixture models to analyze GWAS data involving multiple subgroups. This method reveals the underlying effect size distributions of relevant variants from summary association data to evaluate the existence of variant subgroups and their effect sizes rather than identifying individual variants via single-variant association tests with a conservative genome-wide significance criterion. Here we demonstrate the strength of the newly developed method by applying it to a recently published large-scale GWAS dataset for schizophrenia and bipolar disorder.

In disease-association studies using neuroimaging data, the detection of disease-associations and the estimation of the magnitudes of the associations or effect sizes are essential statistical tasks. In this paper, we propose a framework for modeling effect sizes of whole-brain voxels. Specifically, we assume a semi-parametric hierarchical mixture model to flexibly estimate the underlying effect size distribution for the whole-brain voxels and a hidden Markov random field model to incorporate the spatial dependency among contiguous voxels. We then derive a multiple testing procedure for controlling the false discovery rate (FDR) and a procedure for effect size estimation, for both individual voxels and prespecified regions that have structural or functional features. The proposed method for the estimation of effect sizes reduces the selection bias caused by selecting voxels or regions with the greatest observed effect sizes, possibly due to random errors. An application to neuroimaging data from an Alzheimer’s disease study is provided.

We propose a new gait feature extraction method for gait recognition that uses the dyadic wavelet transform (DYWT) and structural similarity (SSIM). Gait recognition is a person authentication technique that uses the gait features obtained from a monitoring camera video. Most existing gait recognition methods use gait features based on the silhouette image sequence. However, the gait recognition accuracy decreases if the resolution of the silhouette image is low. We developed a new method to extract gait features from silhouette images having a low-resolution gait period sequence that uses the DYWT and SSIM. For our experiment, we prepared two types of silhouette image sequences of 100 subjects for use as the probe and gallery images, respectively. The sizes of the silhouette images were 64 × 44, 32 × 22, and 16 × 11. We describe our proposed method in detail and present our experimental results demonstrating that the extracted gait features are effective.

We propose an automatic detection method for early stage gastric cancer from endoscopic images that uses the fast discrete curvelet transform (FDCT) and wavelet transform modulus maxima (WTMM). First, we convert an original image to CIE L*a*b* color space and apply the dyadic wavelet transform (DYWT) to the a* component of the image. Next, we compute the WTMM using the high frequency components obtained by the DYWT and generate a binary image based on the WTMM. We then produce a composite image using the L* and a* components of the image and compute the curvelet coefficient by applying FDCT to the composite image. In addition, we apply the inverse curvelet transform to only the imaginary component of the curvelet coefficient and generate a binarized image by applying threshold processing to the obtained image. We divide the two binarized images into non-overlapped blocks and compute the average value of each block. The average value of the block corresponding to the abnormal portion tends to be smaller than that of the entire image. On the basis of these features, we describe in detail a method of detecting abnormal regions and show by our experimental results that the proposed method can detect the region in the endoscopic image that may indicate early stage gastric cancer.

The scenario of great competition among companies has awakened the constant need to optimize the way their internal processes are conducted. The continuous process improvement cannot be achieved in any way without strategic alignment, without digital transformation and without the analysis of the human resources responsible for carrying out the activities. The three dimensions of management mentioned above need to be synchronized for the perfect functioning of the organizational gears. This article presents a Business Transformation Methodology that integrates the business process management (BPM) with the strategic objectives, digital transformation and people management to improve the operational efficiency of the business. In addition, an in vivo experiment on the implementation of the methodology using agile project management in a food distributor company is also described.The results of the experiment showed significant improvements in the goods receipt process at a food distributor company, reducing the total execution time of the process, reducing the number of errors and acquiring more control of the critical activities of the company.

Water has become an essential resource to be preserved globally. The need for monitoring the distribution of water with good quality and also meeting the consumer demand is highly essential. This is highly possible by means of Internet of Things (IoT) with edge/fog computing and cloud. Now there is need for intelligence in IoT for monitoring the health of the pipe periodically for proper water distribution thereby meeting the consumer demand. These needs to be automated with minimal human intervention So accordingly, we here propose an multiagent based Smart Water Distribution System employing IoT technologies integrated with Fog for underground pipe health monitoring system. The agent at the aggregator would monitor the underground pipeline health and report to the nodal agent at the Edge. The nodal agent at the edge/fog performs the analytics of the pipe health based on the real-time values and also it predicts the future impact on the health of the pipe using rule-based system on the basis of fluid mechanics. In addition, the agents in cloud is responsible for demand forecasting and pricing calculation for the water consumed. The inter communication between the agents, consumers and SCADA engineers will happen through inter-agent communication mechanism.

Tunneling is a networking approach to virtually encapsulate some channel of private communication within another channel, which is usually public, through means of encryption. Tunneling protocols allow for the establishment of Virtual Private Networks (VPNs) which are useful for anonymity and access to private networks behind firewalls. The many tunneling protocols generally take the form of one networking protocol being transmitted over another, or even over the same protocol. One noticeably less represented variation is TCP-over-TCP due to the overall degradation of performance which is observable as a distinct loss of overall end-to-end throughput of application data, called the goodput. This known loss of performance is a product of the multiple, nested congestion control algorithms inherent to Transmission Control Protocol (TCP) and has been coined the TCP meltdown problem. In this research, we have investigated the contributions that multiple factors play in degrading the goodput of TCP-over-TCP tunnels. Through ns-3 simulation we have studied the performance of the tunnel as we vary transfer buffer size, congestion avoidance algorithm, bandwidth of inner and outer channels, and drop rate. Our simulation is built with the ability to vary those parameters plus more for future cases. Through this analysis, we were able to find the performance of 448 different configurations, not counting experimental control cases.

Intersection Traffic Control (ITC) attracted extensive attention with the increase of traffic accidents and congestion that causes a huge social and economic losses. Also the topic on ITC has been widely studied by research communities. These studies fall into two categories: traffic signal scheduling and Autonomous Intersection management (AIM) based on Intersection Central Unit (ICU). The optimization of signal scheduling needs a lot of computations and the ICU based system is not flexible and not cost efficient. To solve this problem, the distributed AIM system should be the necessary requirement. However to our knowledge there is almost no work that has been devoted to distributed ITC for AV.This paper proposes a token-based group mutual algorithm for the distributed AIM. We use a token as a privilege to pass cross zone (critical section in an intersection) and use one token. How to transfer the token among vehicles in a dynamic system in which a vehicle can not stay in the cross zone after it finish passing it is the core of our algorithm. We design the circulation of the token efficiently among vehicles and session declare concept to improve the flow of vehicles.The proposed algorithm can handle quite a big traffic volume well with decreased message complexity. It outperforms the reference algorithms in message complexity and presents better result in system throughput than traffic signal system. This paper also will vitalize the researches about the distributed AIC.

Cardiorespiratory arrest often occurs in hospital settings. The American Heart Association, the leading institution in treating heart disease worldwide, proposes to treat this condition using the Advanced Cardiovascular Life Support (ACLS) protocol. Despite the existence of this protocol, its procedure is not always correctly followed during the treatment of a cardiorespiratory arrest. This study proposes the development and laboratory testing of a mobile application to aid the cardiopulmonary resuscitation process following the guidelines of the ACLS protocol. The main results showed that the application is a good alternative to support the cardiopulmonary resuscitation process.

The use of openEHR archetypes facilitates the development of flexible and interoperable health applications and allows end users to specify Electronic Health Record (EHR) requirements. However, there are few tools and frameworks available on the market that support health-standard-based Health Information Systems (HIS) development. This work presents a framework capable of creating data schemas for EHR storage and generating archetype-based graphical user interfaces (GUI). A software architecture was developed to create relational and NoSQL data schemas, independent of storage technology, as well as generating GUIs with data persistence capabilities. To evaluate the proposed framework, a health application was developed using archetypes available in the openEHR clinical knowledge manager (CKM). The evaluation results indicate a 72% reduction in coding efforts to develop health applications using the proposed framework.

We investigate the use of the eigenvector centrality (EVC) metric to rank the vertices in a disease-disease network built from the results of the disease-gene association studies reported in the NIH GWAS catalog and OMIM database. The vertices in the disease-disease network are the diseases and there exists an edge between two vertices if the corresponding diseases share at least one gene in the disease-gene association network. The weight of an edge in the disease-disease network is the number of shared genes between the end vertices (diseases) of the edge. The EVC value (ranging from 0 to 1) of a vertex/disease in such an undirected weighted graph comprehensively captures the impact of the number of edges incident on the vertex and the weights of these edges as well as the number of edges incident on the neighbors of the vertex and the weights of these edges. The distribution of the EVC values of the vertices exhibit a Pareto pattern (80-20 rule) such that only about 18% of the diseases have higher and appreciably different EVC values and the remaining 82% of the diseases have lower and comparable EVC values.

In recent years, the rapid aging of the Brazilian population has been verified, making the elderly an emerging age group. Parallel to this transformation, the spread of Information and Communication Technologies, particularly the use of mobile devices has been observed. This fact has raised a great interest in the elderly public regarding the use of these devices, considering the benefits they can provide to their lives. This research aims to evaluate the usability of the Siri virtual assistant, available on mobile iPhone devices, with elderly users. The method used in this research proposed three tasks and usability was investigated from the performance analysis based on time and perceptions of the volunteers through questionnaires. Age was found not to be the only factor influencing the usability of the Siri virtual assistant. Other factors, such as prior knowledge, education level and personal motivation, are capable of influencing the research results of usability criteria.

Demographic data has shown the ageing of the Brazilian population and an increase in the use of the Internet by this group. In parallel, new information technologies and, especially, the use of mobile devices, has multiplied. The elderly public has shown great interest in the use of these devices, considering the benefits and facilities they can provide to their lives. In addition, the number of elderly users who access social networks has increased in recent years. This research aims to evaluate the usability of the Facebook application on mobile devices for elderly users. Age was found not to be the only factor influencing the usability of the Facebook application. Factors such as prior knowledge of the use of smartphones and social networking and personal motivation are capable of influencing the research results of usability criteria.

Ambient Assisted Living (AAL) includes intelligent systems designed to monitor, assist, and promote a healthy environment. These technologies are an excellent opportunity for improving the lives of people, especially those who live with disability, illness, or aging. Much work has been developed for assistance through Ambient Assisted Living technologies, but the majority focus on the elderly, who have different needs from people who are visually impaired. The study presented in this paper explored the needs of people who are visually impaired and requirements for a virtual assistant in an AAL setting. In order to do so, we created an online questionnaire study and obtained 27 respondents who are visually impaired. Results indicated that 81,5% of respondents accept the idea of having a virtual assistant in their home. Also, we were able to gather a number of the difficulties faced by the users in their home, as well as the tasks that must be performed by virtual assistants to support such users.

We investigate the fine-grained object classification problem of determining make, model and year of car from a video. To this end we introduce a new dataset called CarVideos. It is a novel dataset for fine-grained object classification in videos. The CarVideos dataset contains over a million video frames annotated with bounding boxes around the visible cars as well as the specific year, make and model of each car. We implemented several state-of- the-art methods for object classification in videos and compared them using the dataset to establish a baseline performance level for future research. We also introduce a novel approach to fine-grained object classification in videos that combines a Single Shot Multibox Detector (SSD) with a single stream multi-region convolutional neural network (CNN). Our experiments show that our novel method significantly outperforms previous methods in terms of accuracy on the dataset. Our approach outperforms Temporal Segment Networks (TSN) and 3D Convolutional Networks, which are state-of-the-art on human action recognition in videos.

This article brings a proposal to implement a passageway detector for a drone, like windows, doors, holes etc. In this case, it was applied on the model Tello of Ryze Tech. The modeling technique uses a Neural Convolutional Deep Learning Network, with a pre-supervised training. This training is done with three image classes: unobstructed path, obstructed path, and the passageway inside of the path, to a specific environment, such as a school or a hospital. After the detection of a way defined by the user, the technique uses image filters to find a polygon through the window and compare the returned data with the values stored in the neural network dataset. To find the best parameters to identify the passage in the processing, the algorithm makes an adjustment through parameters interpolation that allows the drone to perceive its crossing for many cases of environmental variations. The Network model used is the SSD implemented on Google’s TensorFlow framework and for the image processing, it uses filters and functions from OpenCV library, where both codes are implemented in Python programming language.

This paper has the goal of presenting a deep neural network algorithm for real-time object tracking called Single Shot MultiBox Detector – SSD as a source of detection, in combination with an Optical Character Recognition – OCR algorithm, both serving as assistance for determining the right position and speed for a quadrotor during an active car track mission. During the experiments, a Tello DJI drone equipped with a frontal camera and a distance sensor were used to receive height measurements. The whole algorithm was implemented in Python programming language as a combination of Tello SDK for software development, Hanyazou’s TelloPy package for video streaming, an OCR algorithm implemented with digital image processing and an SSD system created with TensorFlow. The frontal camera was used as a real-time streaming source for the SSD and OCR systems, being both coordinated by TelloPy, which made it possible to achieve both position and speed control using Tello SDK. The distance sensor under the drone helped to avoid collisions underneath and refine the positioning. Outdoor tests were executed to check the drone’s behavior during the active car track mission. Besides the project is still in its initial stage, satisfactory results were collected and will be used for further analysis and improvements.

The user study presented in this paper has investigated the potential benefits of incorporating simulation into the learning process. Simulation-based learning has been employed in the field of science and technology to improve the quality of teaching and learning. How effective is simulation in the enhancement of learning, especially compared with traditional learning techniques, is an important research question that can lead to significant benefits. The work presented in this paper has examined the differences between two instructional techniques: using simulation, and respectively, studying online material. For this purpose, we used a simulation that helps students learn about the rules of ship navigation, with the hypothesis that simulation is better than online study for learning difficult concepts. The study used simulation and online material in a self-study environment, with the research aimed at identifying the best of the two learning methods. The results show that the participants who used simulation during their self-study increased their knowledge from pre-test to post-test. These results also suggest that, compared with traditional learning techniques, integrating simulation into education would be a better approach to enhance learning.

Deep learning (DL) is a process that consists of a set of methods which classifies the raw data into meaningful information fed into the machine. DL performs classification tasks directly from sound, text, and images. One of the famous algorithms for classification of images in DL is convolutional neural networks (CNN). In this research, we tested DL model for image recognition using TensorFlow from Dockers software. We received 99% accurate to identify the test image. The system configuration used for this research includes Ubuntu 16.04, Python 2.7, TensorFlow 1.9, and Google image set (Fatkun Batch Download Image: Google, Google, chrome. google.com/webstore/detail/fatkun-batch-download-ima/nnjjahlikiabnchcpehcpkdeckfgnohf ).

Organizations are increasingly interested in using microblogging platforms, such as Twitter, to get rapid feedback in several domains using sentiment analysis algorithms to rate, for example, whether a target audience is happy or unhappy. However, posts on microblogging platforms can differ from the source material used to train the sentiment analysis tools. For example, emojis and emoticons are increasingly employed in social media to clarify, enhance, or sometimes reverse the sentiment of a post but can be stripped out of a piece of text before it is processed. Responding to this interest, many sentiment analysis algorithms are being made available as web services, but as details of the algorithms used are not always published on the website, comparisons between web services and how well they deal with the peculiarities of microblogging posts can be difficult. To address this, a prototype web application was developed to compare the performance of nine tweet-related sentiment analysis web services and, through targeted hypotheses, to study the effect of emojis and emoticons on polarity classification. Twelve specific research test sets were created with the application, labelled by volunteers, and tested against the analysis web services with evaluation provided by two- and three-class accuracy measures. Distinct differences were found in how the web services used emoticons and emojis in assigning a positive or negative sentiment value to a tweet, with some services seeming to ignore their presence. It was found in general that web services classified polarity sensitive tweets significantly less accurately than tweets where the sentiment of the emoji/emoticon supported the sentiment of the text.

Detecting sarcasm has been a problem in Artificial Intelligence (AI) because it is highly dependent on context and a human agent’s knowledge of the world. This means that for conventional AI, a large number of rules must be hard coded. Furthermore, naïve machine learning methods such as logistic regression simply generate lists of words that are frequently associated and dissociated with sarcasm. Thus, logistic regression is unable to take groupings of words into account in sentiment analysis problems. In this paper, we design a deep neural network that leverages the advantages of convolutional neural networks (CNN) and Long Short-term Memory layers. This CNN-LSTM neural network with word embeddings is then trained 21,709 on word vector encodings of news headlines to determine whether a headline is sarcastic or genuine. We then applied this neural network to a corpus of 5000 test examples of real and sarcastic news headlines. This Deep CNN-LSTM neural network architecture can classify whether a news headline is real or satirical with 86.16% accuracy.

Nowadays telecommunications industry is based on closed source software which prevents the developers to be innovative (which blocks the innovation of developers). Distributed and open sourced telecommunication systems coupled with blockchain technology may have the ability to prevent errors and provide integrated user experience. Failure to pay sufficient attention to privacy; a need for a third party to verify transactions; high costs of services and long transaction times has made current communication systems vulnerable and therefore reducing the satisfaction of customers. This situation forces businesses in the telecom industry to consider some technological measures. The aim of this research is prioritizing the capabilities of blockchain in telecommunication industry for promoting customer satisfaction. First the applications of this technology are introduced. Second, the criteria for telecommunication are extracted from literature and then 3 important criteria are selected in an expert panel with 6 experts. The target populations are the customers that receive the services of telecommunication industry businesses. In the next step, customer satisfactions of blockchain applications in the telecommunications industry are ranked by Analytic Hierarchy Process (AHP). The results show that data security and privacy are the top priorities for customers. The current research is conducted in a theoretical manner in Iran in 2018, so and the results of this research can be useful for the managers of telecommunication industry businesses.

In this paper we use the idea of context-based orthographic error corrections by taking the TREC-5 Confusion Track Data Set’s degrade5 and attempting to correct errors generated during Optical Character Recognition. We do this by identifying all errors using OCRSpell, then generate the 3-gram and searching for the first and last word in the Google Web 1T corpus of trigrams. We then select the candidates with the highest frequencies and a small Levenshtein edit distance. We report on our accuracy and precision and discuss on special situations and how to improve performance. All source code is publicly available for our readers to further our work or critique it.

In order for students to be better served; a system that can be designed to adapt education delivery to students’ needs as well as give students an immersive education experience is needed. Classes are often comprised of students with diverse learning styles and differing proficiency backgrounds and as a result, the traditional education system often under-serves many students. Game Based Learning (GBL) includes applications and software used to train, teach, or facilitate the learning of a subject. We proposed a game called Ecomerica, to show how a GBL system can lead to better learning of introductory topics on economics.

The student Admission System described in this paper was developed and implemented in-house for the Admission department of the higher education institution in Kazakhstan. It is one of the key information systems and is part of an integrated student information system environment, which is intended to automate various areas and processes of a university and the student lifecycle. This paper shows the features, components and the architecture of the developed information system and its framework. The system has the architecture of a web application. The main stakeholders and users of the system are the employees of the Admission Department and potential students/applicants of the university. The system was developed and enhanced during the last few years with such technologies like IBM WebSphere Portal, IBM HTTP Server 7.0.0.31, IBM Forms 8.0.1, IBM Tivoli Directory Server 6.2.0.0, Oracle 11 g Database, Red Hat Linux, Java and JavaScript. The proposed software architecture and experience could be followed and used by various development teams, who plan to develop local web-based applications for universities, for automating business processes of an Admission Department.

During the past decade, the smart grid (SG) concept has been advancing better global acceptance of the electric vehicle (EV) notion. Additionally, the incorporation of renewable energy (RE) resources in the SG is a promising solution to reduce carbon emissions in an intelligent manner. Due to technological barriers at this moment, battery charging time is one of the most significant hurdles for wide adaptation of EVs. The average charging time for EVs is relatively long compared to conventional vehicles refueling. Battery Exchange Stations (BES), which can offer battery exchange plans, facilitate the adaptation of EVs into the SG. Optimizing the incorporation of components such as EVs, RE, and BESs into the base of the SG is a key challenge. In the literature, there are many studies on BESs, which propose algorithms in a solus manner to improve the incorporation of existing components into the SG. However, none of these studies have proposed a comprehensive model that considers the incorporation and transaction of all of the above mentioned components into the SG. In this paper, the authors present a survey on the existing algorithmic approaches to adapt EVs into the SG through BESs. They then introduce the concept of a battery consolidation system, which is a solution that focuses on optimizing the incorporation and transaction of all of the components in the SG. The authors propose a system model for a battery consolidation system, and then formulate the BES optimization problem based on the system model. Finally, the authors present different scenarios for some theoretical situations for EV adaptation in the SG.

Harvesting high output power from ambient vibration energy using a hybrid piezoelectric and electromagnetic equivalent circuit is a verified technique. This paper introduces a novel interface circuit for the hybrid system, which has high efficiency and output power. The proposed interface uses a parallel-synchronized switch in the standard AC-DC converter for the piezoelectric energy harvester part, which can greatly improve the output power and efficiency. Moreover, a DC-DC boost converter is used to enhance the extracted energy from the electromagnetic energy harvesting section due to its low output power. The defined interface model implemented on a typical hybrid piezoelectric-electromagnetic system and the simulation results confirm the enhancement of output power to 250 mW along with the efficiency of 80%. The efficiency of the proposed hybrid harvester enhanced 47.36% and 92% in comparison to the standard hybrid and piezoelectric system respectively. The effectiveness of the hybrid circuit confirmed while its extracted power is 50 mW more than the single piezoelectric system with the switch interface.

Power-down mechanisms are well known and are widely used to save energy; these mechanisms are encountered on an everyday basis. We consider a device which has a set of states defined by a continuous function. The state of the device can be switched at any time. In any state the device consumes energy and a power up cost to switch the machine to full capacity. This project gives experimental results regarding power consumption to satisfy service request based on online competitive analysis. Competitive ratios, which show the effectiveness of the algorithms compare to the optimal solution.

We review existing algorithms for the placement of towers for illuminating 1.5D and 2.5D terrains. Finding the minimum number of towers of zero height to illuminate 1.5D terrain is known to be NP-Hard. We present an efficient algorithm for solving a variation of the tower placement problem in which we are required to place a tower of given height to maximize the coverage. The main ingredient of the proposed algorithm is based on discretizing the problem domain by using a novel method for identifying feasible placement points.

Sentiment analysis is a series of methods, techniques, and tools about detecting and extracting subjective information, such as opinion and attitudes, from language. The goal of our project was to classify movies’ reviews, by analyzing the polarity (positive or negative) of each paragraph in a review (Cui et al., Neurocomputing 187:126–132, 2016). We experimented with various RNN models on the Nervana Neon deep learning framework, an open-source framework developed by Nervana Systems, in order to improve accuracy in training and validation data. We experimented with network architecture, hyper parameters (batch size, number of epochs, learning rate, batch normalization, depth, vocabulary size) in order to find out which model works best for sentiment classification. This paper presents our findings and conclusions.

The demand for low power processing is increasing due to mobile and portable devices. Pipelining is extensively use to improve the throughput of the processing unit. However, to implement a pipeline requires adding a register at each sub-stage that results in increasing the latency. In a processor, an adder is an important building block since it is used in Floating Point Units (FPU) and Arithmetic Logic Units (ALU). Therefore, for improving the overall performance of a computing processor, designing a low power pipeline adder with low latency is required.In pipeline architecture Dual Edge Triggered (DET) based register can help in reducing the latency. However, for high input activity, a DET flip-flop consumes more power than a Single-Edge Triggered (SET) flip-flop. Additionally, replacing each Flip-Flop (FF) in the processor with Dual Edge Triggered (DET) has a considerable area and power overhead.In this paper, we are proposing a new shift register which imitates DET FF based shift register without the need of special DET FF. The proposed shift register improves the latency in a 4-bit pipelined adder by twofold. Additionally, the power delay product was reduced by 44.16%.

The maximum clique problem is to find the largest set of pairwise adjacent vertices in a graph. The problem has been shown to be N P $$\mathcal {N}\mathcal {P}$$ -hard. This paper presents an approach to solve the maximum clique problem based on a constructive genetic algorithm that uses a combination of deterministic and stochastic moves. The problem has wide applications in areas such as bioinformatics, experimental analysis, information retrieval, signal transmission, and computer vision. The algorithm was implemented and tested on DIMACS benchmarks. Favorable results are reported.

Context: Microservice-oriented architecture relies on the implementation and deployment of small and autonomous microservices, rather than implementing the functionalities in one unique module to be deployed. They have been adopted as a solution to the shortcomings of the monolithic architecture such as lack of flexibility. Goal: This paper discusses lessons learned and challenges reported in the literature regarding the migration of legacy monolithic software systems to microservices based architecture. Method: We performed an automated search targeting public repositories to accomplish the stated goal. Results: Based on the evidence provided by 12 studies, we classified main findings in lessons learned related to the migration, as well as associated difficulties and challenges. Conclusions: the guidelines to migrate to microservices based architecture are maturing/evolving and the literature has pinpointed issues that deserve further investigation.

The rise of Quantum Computing in the industry begets a closer analysis of the topic and its methods. Quantum computers are not limited to the two states 0 and 1; rather, they encode information as quantum bits, or qubits, which can exist in superposition and can be entangled. Understanding the entanglement of qubits requires an understanding of the Bell states. The first Bell state(Φ +), has been widely studied and is prominent in quantum computing literature. However, the other three (Φ −, Ψ+, Ψ −), are relatively untouched by researchers since Quantum Computers have only recently been available for research that could test them. Years of theoretical research has not been backed by experiments due to the lack of available technology. Now with its availability from companies like IBM we have the resources to test and work with all four Bell states, further developing Quantum Computing as a strong pillar in computing.

This paper proposes the selection of historical cost performance data of processes using the Random Tree classifier and the application of the calculation proposed by [20], to improve the predictability of projects cost. The proposed technique was evaluated through an empirical study, which evaluated the implementation of the proposed technique in 23 software development projects. The proposed technique has been applied in real projects with the aim of evaluating the accuracy and variation of the CPI Accum and consequently the EAC. Then it was compared to the EVM traditional technique. Hypotheses tests with 95% significance level were performed, and the proposed technique was more accurate and more stable (less variation) than the traditional technique for calculating the Cost Performance Index – CPI.

Currently the Project Management Institute (PMI) estimates that approximately 25% of the world’s Gross Domestic Product (GDP) is spent on projects of various kinds and that about 16.5 million professionals are directly involved in project management worldwide. This volume of projects and changes in the world scenario, increasingly competitive, generate the need for faster results, with higher quality, lower costs and shorter deadlines. Among the main techniques for analyzing cost, time and scope performance, the Earned Value Management (EVM) technique is considered to be the most reliable. Several formulas derived from EVM’s measurements are available and have been studied over the past 15 years. However, EVM has a significant limitation regarding quality in its method. The technique is effective in providing cost and schedule related information but is still weak in taking the quality factor into account. The objective of this work is to improve the predictability of cost and schedule of software projects using the EVM technique, by adding quality variables and allowing EVM to integrate not only scope, schedule and cost but quality as well. This paper presents a proposal to enhance the EVM technique by integrating the quality component. The proposed technique is evaluated and compared to the traditional technique through different hypothesis tests, utilizing data from simulated projects. Hypotheses tests with 95% significance level were performed, and the technique was more accurate than the traditional EVM for the projection of the Cost Performance Index – CPI and the Schedule Performance Index – SPI.

This article examines the main published works on predictability of cryptocurrencies from a point of view of identifying techniques, algorithms and characteristics that can improve price predictability in cryptocurrencies. This paper presents a systematic mapping of papers and classifies them into categories on type of analysis and techniques. This work can help the scientific community to better identify and structure a fairly emerging area that is the cryptocurrencies and how their markets can be studied.

During the first semester of 2017, at the Brazilian Aeronautics Institute of Technology (Instituto Tecnologico de Aeronautica – ITA), 30 undergraduate and graduate students from three different courses applied Interdisciplinary Learning Based on Problems (IPBL) and during 17 academic weeks had the opportunity to conceive, model and develop a Computer System to combine data and integrate different actors such as PATIENT, HOSPITAL, PHYSICIAN and SUPPLIER to the decision-making process related to health crisis management, such as epidemics. The purpose of this system based on Big Data and Internet of Things (IoT) was to manage data and information to allow the appropriate decision making. The agile Scrum method and its best practices with Python or Java, Spark, NoSQL databases, Kafka and other technologies were applied collaboratively in a fictional crisis scenario as a proof of concept (PoC) to solve health system problems. The main contribution of this study was the use of agile tests in the verification and validation of this academic system related to the management of health crises.

This paper examines Kant’s categorical imperative as an ethical theory that is essential and important in today’s organization. Literature has discussed the importance of Kant’s categorical imperative in regard to the ability to set a moral vision and conduct that will be acceptable and universal to follow. This paper further explains the content of Kant’s categorical imperative and how it relates to business ethics in the field of project management in aerospace and defense industry. Finally, this paper discusses the application of Kant’s categorical imperative that can be used to design an ethics program in an organization.

In countries with very extensive territory, such as Brazil, we have some problems such as monitoring the deforestation of the Amazon forest or monitoring the country’s border. A solution to lower the cost of this monitoring would be the use of RPAs (remotely piloted aircraft), but the batteries still have very low autonomy, thus failing to cover a large area. To try to solve this problem, the idea is to take advantage of the solar rays, using solar plates to recharge the battery during the flight and using the benefits of the upward currents where they can be used to save the battery, limiting the use of the motors to onlu in the most necessary parts of the flight.

Mosquito-borne diseases negatively affect economically emerging countries. Nevertheless, the current public healthcare solutions are insufficient to support disease surveillance and control. The citizen engagement in reporting mosquito breeding sites is hard to achieve but essential in preventing disease outbreaks. This paper introduces the VazaZika platform aimed to support the surveillance and control of mosquito-borne diseases. This platform evolves the VazaDengue legacy platform with gamification. Through game elements and rules, we aim to make enjoyable and challenging to report mosquito breeding sites via VazaZika. Citizens are continuously rewarded as they perform tasks in the platform. They progress in levels that enable new tasks and jump in rankings according to the citizens’ location. Citizens can also join teams for engaging with challenges, which helps to develop a sense of belonging and connection against the spread of diseases. This paper reports the process of gamifying VazaDengue, the platform user interface and its conceptual model, aimed to support reuse.

The Viola-Jones and HOG methods for face detection, and convolutional neural networks for face recognition are used to implement the developed software system. Algorithm testing was conducted. The accuracy of image recognition by these algorithms is estimated. The software system allows you to select any of these algorithms to solve the detection and recognition problem.

In this era of globalization, Higher Education (HE) institutes are required to adapt and provide diverse delivery options in order to sustain the highly competitive international marketplace. With the recent developments in the digital world and the growth of transnational education (TNE), HE institutes are deploying relatively new delivery methods and techniques such as Massive Open Online Courses (MOOCs) as well as collaborative degree-apprenticeships to attract students with different learning styles. This research explores the TNE landscape in general as well as in the context of Australia. We propose flexible TNE delivery models and critically compare them based on key influencing factors for sustainability and future expansions. The paper concludes making recommendations for TNE adoption both onshore and offshore.

Mobile Ad Hoc Networks (MANETS) are highly vulnerable to security attacks than wired network. In this paper, we propose an acknowledgment and signature based intrusion detection system for securing data packets from malicious nodes during packet transmission over MANETS. The system is named as Malicious Node Detection System (MaDS). We implemented and evaluated MaDS in NS2 and compare the result against 2ACK and EAACK in terms of packet delivery ratio.