Search

The trust assessment is a key function that should be performed in advance of any cloud customers’ decision making. A complete methodology for security, risk, and reputation assessment of cloud identity providers is a gap in the current literature and system architectures. In this paper multiple criteria decision-making (MCDM) is introduced to prioritize the attributes for a cloud identity trust framework. The overall trust assessment is decomposed into two parts: the trust analysis of the federated identity management systems, and the quantification of trust. The MCDM approach delivers the inputs to an analytic hierarchy process (AHP) to complete the trust assessment. This paper innovates a theoretical solution to the trust gap between cloud identity providers and the cloud identity customers in the current literature.

Atmospheric Chemistry is important in public health. This paper highlights methodology aspects from analysis of atmospheric chemistry data in China, its correlation with cancer morbidity data, and impact on future cancer morbidity rate that are due to changes in climate.Forthcoming papers use DL for prediction. Research on cancer and analysis of cancer morbidity are provided by Liu Yuxin, Zhaorong Zhu [5].

This paper is about mapping innovations in Artificial Intelligence (AI) with the help of patents. It leverages the use of Social Data Science techniques on two levels: (1) we perform a content analysis of all the abstracts of the available patents of an extensive database; (2) we use a Latent Dirichlet Allocation (LDA) technique on these patents’ abstracts to extract which categories best describe each subfield of AI. Our goal is to have a deeper perspective and a deeper understanding of the various developments in AI through time and geography. The database used in this paper is one of the most comprehensive, with a total of 55,109 patents. To contextualize this study, the literature review uses information from 29,225 articles. In both cases, the analysis of such amount of information could not be possible without dedicated computing power.

The Internet of Things has recently gained extraordinary popularity. This popularity is only expected to grow in the coming years. Being able to receive services and information from Things anywhere in the world will bring big changes in every facet of people's daily lives. Things are considered service providers themselves. A user should be able to find a Thing and request the services provided by Things. But what happens when the number of Things become so large? How will the user be able to find and select a Things? How will users select the most appropriate and beneficial Thing to carry out a requested task? All of these are some of challenges that are rising from the popularity of the Internet of Things. This paper will investigate the above questions and try to solve them using two well established software engineering concepts: service-oriented architecture and context-awareness. The main contribution of this paper is a Service-oriented Context-aware Architecture for the publication, discovery and provision of services in the Internet of Things.

During the past decades, face detection has attracted significant attention. Many papers have been published utilizing various methods for face detection. One of the most popular face detectors used in many practical applications is Viola-Jones method. Despite of being a real-time and robust face detector, it suffers from not well explained parts at the training procedure. Such as, "not clear how to select few features in the first cascades" or "not clear how many samples are needed and how to gather a good train set for training a cascade". In this paper, a train set selection method based on histograms generated from AdaBoost and also, a simple method to select few features in beginning cascades are proposed. The training procedure is then compared to a baseline training presented in the previous works.

This paper discusses the sensor technologies used in city and university settings for smart parking systems. In high density cities, parking can become time consuming and result in increased traffic congestion as drivers find parking. Currently, most drivers in these situations attempt to find parking without aid. Knowing in advance if parking would be available provides a valuable feature in a Smart City. While smart parking solutions have already been researched, this paper discusses current implementations and sensor technologies used and compares the benefits and drawbacks of the main approaches in literature. Presenting parking information to users is also a vital part of a smart parking system. Finally, we will discuss the details of a proposed Smart Parking System - SParkSys.

Melanoma is the most deadly form of skin cancer and its incidence rate is significantly increasing. The design of an assisted diagnosis system for the detection of melanoma is a challenging task involving various steps related to computer vision. Researchers have concluded that the accurate identification of melanoma requires robust preprocessing steps on dermoscopy images including hair removal, illumination correction etc., that can help in a better detection of melanoma. In this paper, we propose a novel illumination correction algorithm followed by robust feature extraction from dermoscopy images, leading to a better identification of cancer. Illumination correction is based on statistical estimation of illumination content in the images, followed by the extraction of differential structures using a combination of Gabor filtering followed by extracting local mesh patterns, which exhibit physiological significance based on various clinical rules for detecting melanoma. Our experiments show that the proposed technique outperforms all the other methods that have been considered in this paper.

There have been significant advances in computer graphics and computer vision for description and recognition of rigid shapes and objects. However, the problem of description and recognition of non-rigid shapes and objects is still an open problem. In an earlier study \cite{IEEEhowto:MuNeArBh}, we proposed a geodesic field space-based approach to describe and analyze non-rigid shapes from a point correspondence perspective. In this current study, we describe a canonical set of shapes, very much like a dictionary with mathematical structures that are inherently relevant to the shape classification task. We propose a novel method of representing 2D shapes where every planar shape will be assigned a unique fingerprint, which is a conformal map of the shape to a canonical shape in the plan. We believe that the study presented in this paper can be extended to describe and recognize both, rigid and non-rigid shapes and objects. The main focus of this paper is to provide an insight into the mathematical and geometrical fundamentals. We will restrict our attention to the case of doubly-connected planar domains.

Robot Audition enables a robot to listen to simultaneous utterances via its own ears by localizing and separating sound sources and recognizing separated sounds. Although several applications are developed as a proof of concept, such capabilities of robot audition have not been well shaped in the context of applications, particularly, in multi-parson interactions. This paper focuses on the question ``what is the next step when a robot and/or system can listen to several utterances at once?'' As an example of multi-party interactions, the paper describes a quizmaster robot for speech-based Jeopardy! like games with two interaction models, school-class-type and auction-type. The player of the quiz answers a question by getting the right to answer in the former, while he/she can say an answer directly. Empirical evaluation of the system and lessons are discussed.

A lack of engagement by students has been found to be a factor in the high attrition rates in Information and Communication Technology (ICT) courses. This paper explores whether using a flipped-classroom approach and continuous assessment can increase student engagement in first year in their learning of ICT professional skills and ethics – a topic that previous research has identified as a low interest topic. A case study research methodology is used to demonstrate that increasing student engagement could impact positively on both attrition and academic performance.

Artificial neural networks have gained prominence by being used in different aspects such as medical diagnosis and detecting malware applications. However, neural network models could have an error rate that indicates their performance. Thus, optimization algorithms can minimize the error rate by updating the neural network parameters to reach an optimal solution. This paper explores the use of permissions and underlying Linux system information features in Android platform to evaluate gradient descent optimization algorithms in neural networks. Those optimizers are evaluated by running them on a set of Android applications to find the optimum one. Furthermore, each optimizer is assessed based on its default and adjusted parameters values. This evaluation shows that the best accuracy score is 92.21% collected by Adam optimizer.

In this paper, we propose a model-based methodology to evaluate the impact of the wireless network conditions on the robustness of performance of adaptive and non-adaptive networked mobile applications (apps). The methodology consists of three steps and it requires three different artefacts as inputs: the network model, the app behaviour model, and the performance model. To quantify robustness, two metrics (static and dynamic robustness) are proposed. The main challenge in evaluating robustness is the combinatorial growth of network-app interactions that need to be evaluated. To mitigate this issue, we propose an algorithm to limit the number of interactions, utilizing the monotonicity property of the performance model. To evaluate the dynamic robustness metric, the ability of the adaptive app to tolerate degraded network conditions has to be evaluated. This problem is formulated as a minimization problem. The methodology is used to evaluate the performance robustness of a mobile multimedia streaming app. The effectiveness of the proposed methodology is evaluated. The obtained results show three to five times reduction in total cost compared to the naive approach in which all combinations are exhaustively evaluated.

This paper presents the design of UWB flexible wireless antenna for WLAN, 5G, and WiMAX applications. The design is a pentagonal shaped radiating element placed on top of flexible polyimide substrate and simulated using the wellknown 3-D electromagnetic (EM) simulator HFSS, v.18.1. As the EM simulator required large computing cluster to solve the design under consideration in addition to the time consumed, Artificial Neural Network (ANN) is used to implement the design and eliminate the factor of time and cost for the aforementioned structure. ANN with 1 hidden layer of 15 neurons based on Levenberg-Marquardt algorithm is presented. An error of less 6% is produced during the leaning, validation, and testing processes. The implemented ANN is a good candidate to represent the design and eliminate the factor of simulation time in addition to the cost of large computing cluster.

Improving the dependability of distributed applications is a challenging task. Distributed predicates detection techniques (runtime verification) can help a lot in this regard by verifying that a given run of a distributed application satisfies certain properties (formally modeled as predicates). In general, these techniques can be exploited in the verification of the correctness of a particular implementation of a distributed system. In any distributed application, several processes are usually running concurrently. Consequently, the number of global states to be considered by the detection techniques is exponential (NP-Complete problem). In this paper, we have investigated the effectiveness of using a meta-heuristic algorithm in solving the problem of detecting distributed predicates. Namely, we have used the simulated annealing (SA) algorithm to develop an algorithm to detect distributed predicates. The results of the experiments demonstrate that the SA-based detection technique outperforms the state enumeration based detection technique.

A con man trick appears in cyberspace as a deception around cloud services from the service provider. Con man resistant trust algorithm continuously monitors cloud service's performance to detect this con man behavior. A cloud service is identified as a con when a lapse in service performance repeats by an amount more than the consumer of this service can accept. Like the repetition, performance of a degraded amount is another service performance attribute affecting consumer satisfaction. The work presented in this paper includes performance degradation following various models. The algorithm's mathematical scheme tracks the performance degradation amount. This revision evaluates con behavior in proportion to the performance degradation. This proportional affect follows an exponential function. Three alternative exponential functions are evaluated. Each of these functions is applicable for specific cloud services. These functions are compared with each other and their properties effecting con resistant Trust are analyzed resulting in a set of principles. Also, the relationship between performance degradation amounts and con behavior detection speed are unlocked resulting in a principle. The result correctness and effectiveness of the principles.

In this paper, an algorithm for auto adaptive edge detection is proposed. A multi-resolution method called Spectral Intrinsic Decomposition (SID) is first used to perform a signal decomposition. The edge detection method is based on thresholding a sequence of filter parameter. The algorithm includes three steps processing: image decomposition, image recomposition with some components and extraction of local maxima greater than a given threshold. The proposed edge detector is adaptive, robust and extract straightforwardly key components. The effectiveness of this proposed method is illustrated with some results on a set of images samples and compared with previous work. The results show that this new method can outperform other existing edge detection methods and can be used for many applications in various fields like in medical image processing.

Internet of Things enables real time monitoring of environmental data. There is an opportunity to collect data in remote areas where real time data was previously unavailable. Ideally we would like to both collect and communicate that data in real time. However communication over GSM networks relies upon power-intensive cell radios which severely limits the ability to deploy sensors for periods of time over one week. In this paper, we discuss an approach to unattended long-term deployment of an environmental sensor using GSM networks to communicate data in real time. We propose a custom power circuit that reduces power consumption rather than depend upon the low power modes of MCU. This approach greatly extends the time a sensor can be deployed with a commodity battery without manual intervention. We present both the circuit design and its evaluation.

Heterogeneous processors based on functionally diverse devices (GPP, GPU, FPGA, ASIC and DSP) may provide affinity for different types of big data processing. In this paper, we developed affinity characterization of such functionally diverse device based heterogeneous systems. However, extracting such characterization is computationally intensive and rather slow for large set of nodes. We first developed general affinity mapping approach which showed 2 to 15 speed up and reduction of energy from 30-35%. We investigated the machine learning approach to the characterization by evaluating predictive performance using training data from disparate heterogeneous computing systems and achieve an improvement in affinity matching. We can extend this machine learning approach to develop comprehensive affinity between big data applications across several domains to heterogeneous computing using functionally diversified processors.

Fingertips positions play an important role in Human Computer Interaction. Traditional methods for fingertip detection, based on RGB images, are designed for stretched fingers only. Recent research, based on only depth images, could not perform well. In this paper, we propose a method for detecting and tracking the 3D position of a single fingertip, and recognizing its motion. Our algorithm is a fusion of 2D and depth detection. In order to promise the accuracy of detection and smooth the movement trajectory, a 3D Kalman filter is proposed. Finally, the hand motion is recognized based on the 3D coordinate sequences of the fingertips. Experiment results have demonstrated that the proposed method performs well because it can achieve an accuracy of 92.86% for fingertip detection.

Cloud networks are kept evaluating and evolving in recent times due to their performance and efficiency, deployment critics, and costs. Recent progress has been quested for more efficiency and high scalability of cloud networks. The improvement over geographic distribution of cloud facilities in performance and resilience has already been supported and produced in the cloud network overlay protocols and advanced IaaS services, which are more reliable and high-performance. This paper gives an overview of cloud networks and distributed services as well as a summary of some cloud network overlay protocols and datacenter distribution features, such as virtual network segmentation, incremental deployment, multipath and load balancing, and multicast.

Data centers require advanced models to study confrontation of big data. We have developed an improved consistent model that processes the incoming tasks in minimal time and in an efficient manner considering two problem statements for this paper: 1) Portfolio pricing and 2) Counting words in dictionary. The exclusiveness of the idea is not to submit the task data with the task. We keep all data separately in a data grid (GD) domain. The main component in our study for this task, job dispatcher (JD), divides the input job in an efficient manner and distributes them to computing grid (CG) nodes, which are responsible for processing the tasks, and sending the result status back to JD and putting the results in the data grid. The job dispatcher aggregates these results.

Grey model GM(1,1) which implies first-order derivative and one independent variable is considered to predict results producing better accuracy. Solving the grey model GM(n,m) for prediction is still being a challenge due to erroneous results of higher order derivatives, where n is the derivative order and m, number of independent variables. In this paper, we solve this challenge through reduction method and recursive computation. Thereupon, we propose a new method called GM(n,m,k) which is generalized using Machine Learning, where k is a set of constraints belonging to m dependent or independent variables of a given system. We have applied and tested the proposed grey model to applications in airlines industry and it is proved to give results with much better accuracy than the traditional grey model.

Watermarking techniques for relational data have been proposed to allow copyright protection and data authenticity, among other things. Almost all proposals depend on the primary key of the database relations for deciding where and how to place the marks. The primary key could be easily updated or deleted if the attacker does not require watermarked data to be placed back in the database. The few techniques trying to avoid this dependency create a virtual primary key through schemes that frequently compute non-unique values, which may cause watermark synchronization problems. Also, the deletion of attributes compromises obtaining same values for the virtual primary key used for the mark embedding. In this paper, we propose a solution that solves the dependency of the primary key, avoiding the problem of non-unique values and showing more resilience against attribute deletion attacks than previous schemes.

Internet of Things (IoT) is basically a system of inter-related computing devices on a network which perform data collection and exchange of real world properties. The major developments witnessed in IoT are mainly within companies, industries, a home environment etc. To increase the scalability of IoT, various security concerns need to be overcome. IoT devices are mostly resource constrained. Complicated network produces potential vulnerabilities referred to heterogeneous devices, sensors and backend systems. So, to realize the dream of internet of things, security is to be ensured for device to device communication. Though IoT has security mechanisms at the network and device level through identification management, it is safer to have a per-service level specific access control based on the “Roles of Things” to attribute for confidentiality and Integrity. This paper presents the current security practices and puts forth the future scope of improvement in fine tuning service level access control.

Outlier detection refers to detecting patterns that do not conform to an established normal behavior in a given data set and has found immense use in a wide variety of practical domains. Due to its important applications, many techniques have been developed for it, including distribution-based outlier detection methods, distance-based outlier detection methods, density-based outlier detection methods and clustering-based outlier detection methods, etc. However, no methods is completely satisfactory. Motivated by the centroid concept of k-means clustering algorithms, in this paper, we propose a k-nearest-neighbor centroid-based outlier detection method that is based on a global outlier factor and a local outlier factor and can provide competing performances with existing solutions. Experiments performed on both synthetic and real datasets demonstrate the efficacy of our method.

Internet of Things (IoT) technologies have gained widespread popularity in recent years. However, the application of IoT technologies in construction and civil engineering is relatively unexplored. This paper describes an IoT monitoring system for the purpose of improving safety in these fields. Validation tests for the system have been conducted in actual construction and civil engineering sites. It is expected that its adoption in the construction and civil engineering industries can greatly enhance safety in the long run.

Density-based clustering algorithms are well known for identifying clusters possessing very different local densities and existing in different regions of data space. However, the parameters required by most popular density-based clustering algorithms, such as DBSCAN, are hard to determine but have significant impacts on the clustering results. In this paper, we present a new density-based clustering algorithm in which the selection of appropriate parameters is less difficult but more meaningful. Experiments performed on several datasets show the effectiveness of our approach.

In this paper, a new multivariate discretization algorithm called multiCAIM is presented. Discretization consists of transforming continuous attributes into discrete ones. Most discretization algorithms are univariate and find a discretization scheme using only a discretization criterion. On the other hand, the proposed approach obtains a set of discretization schemes guiding the search by using a discretization criterion and the prediction accuracy of Naive Bayes. The obtained datasets using multiCAIM are evaluated employing three popular and competitive classifiers: Naive Bayes, KNN and C4.5. The overall assessment suggests that the proposed algorithm outperforms classical discretizers, moreover, it allows to the expert to select more than one discretization scheme.

This paper presents a multi-camera surveillance system that detects and tracks a moving object in indoor and out-door environments. Since each camera has specific field of view (FOV), the problem is how to keep track of the object when it is out of the FOV of a camera. The system uses landmarks to coordinate the tracking tasks among the cameras. In this work we proposed and developed an algorithm to switch from one camera to another, according to the current location of the object of interest. The algorithm attempts to switch to the nearest camera which can observe the object of interest. Experimental test results showed that the system can coordinate the camera activities and track objects of interest as they move in the monitored area.

In order to solve the inadaptability of traditional software and utilize the computing resources of edge nodes in the future Edge IOT Era, the paper analyzes the characteristics of IoT software and puts forward a new distributed knowledge framework. In this study, we introduce knowledge base into the normal wireless node and translate Wireless Sensor and Actor Network (WSAN) topology along with the software logic into knowledge for it. Moreover, seven heuristic rules are proposed in the process of design, based on experience, including node selection criterion and node collaboration mechanism. Finally, the proposed framework is illustrated and evaluated with a small greenhouse scenario.

As the reality of global warming becomes more severe on our environment, policy responses to reduce carbon emission have been created, debated and even deployed in some areas. Although Africans are not directly responsible for the severity of the climate change today, it has become necessary for the continent to join the campaign to save the planet from global warming. It has been debated that these new environmental policies may slowdown economic prosperity in developing nations. However, there is insufficient empirical evidence to back this assertion. The objective of this study is to investigate the relationship between air pollution and economic growth in Nigeria. Secondary data was obtained predominantly from World Bank World Development Indicators. An IBM Watson Analytics approach was used in analyzing the data. The result showed that air pollution variables had both negative and positive influence on the economy. A smart city framework was developed to guide measurement and analysis of air pollution data for informed decision making. The paper recommends that air quality sensors be deployed to all possible pollution generation machineries for real time data generation, analysis and policy response to control pollution.

Recognizing sensitive events in images, such as flood events, is significant for the maintenance of normal public opinion and social stability. By now, it is still a challenging problem. In this paper, we propose a novel method for recognizing flood events using semantics and CNN-based multi-label image classification. Our method utilizes the empirical information of people so as to improve the recognition result. We first define the semantic model of flood event and introduce it into a multi-label classification model which is based on CNN, and then use the event discrimination model to predict the results. Concretely, for the multi-label classification, we propose a novel end-to-end model, which takes the images as inputs and outputs the predictions of multi-labels. The model can be pre-trained with a single-label dataset, and no ground-truth bounding boxes are required for training. The experimental results demonstrate that our proposed flood event recognition method is superior to others.

In the early 1990s, unmanned aerial vehicles (UAV) were solely a domain of the military use by various developed countries. Now, the ease of availability and affordability in the electronic device market, this aerial vehicular technology has augmented its familiarity in public. However, expanded use of UAVs, colloquially known as drones is raising understandable security perturbs. Because of their abilities to get close-in to potential targets, drones are thought to present a threat and, therefore, the investigation of crimes committed by UAVs is a much-needed facet currently. This motivated us to devise a forensics framework proficient enough of examining the drone's activities after its flight. In particular, this paper analyzes the essential major log parameters of the autonomous drone and proposes a comprehensive drone-forensics related software architecture with preliminary results. Our under-development software will provide a user-friendly graphical user interface (GUI) to allow users extract and examine the on-board flight information. This would provide the forensic science community with a tool for investigating drone-related crime cases.

The address space randomization technique was proposed to make determining the address of a shared library more difficult since each instance of the program is loaded into a random base address. However, when address space randomization layout (ASLR) is implemented on a 32-bit system, an attacker can use a brute force attack to guess the address of the shared library. The main goal of the research described in this paper is to study the use of a dispatching algorithm and multiple back-end servers as a moving target defense technique to mitigate ASLR weaknesses. First, we present a brute force attack when the number of servers is known. Second, we present a brute force attack when the number of servers is unknown. Last, we present the probability of the attacker's success on both of the attacks.

In ad-hoc networks, data messages are transmitted from a source wireless node to a destination one along a wireless multihop transmission route consisting of a sequence of intermediate wireless nodes. Each intermediate wireless node forwards data messages to its next-hop wireless node. Here, a wireless signal carrying the data message is broadcasted by using an omni directional antenna and it is not difficult for an eavesdropper wireless node to overhear the wireless signal to get the data message. Some researches show that it is useful to transmit a noise wireless signal which collides to the data message wireless signal in order for interfering the overhearing. However, some special devices such as directional antennas and/or high computation power for complicated signal processing are required. For wireless multihop networks with huge number of wireless nodes, small and cheap wireless nodes without such special devices are mandatory for construction of the network. This paper proposes a novel method for interfering the overhearing by the eavesdropper wireless nodes by a routing protocol and a data message transmission protocol with cooperative noise signal transmissions by 1-hop and 2-hop neighbor wireless nodes of each intermediate wireless node.

In Software Product Line Engineering (SPLE), in the problem space, variability in a product family is specified in an enumerative manner (by a feature model), i.e. all valid variants are enumerated. However, in the solution space, current SPLEapproaches use parametric variability (variability parameterised on features occurring in a single product variant) instead. In this paper, we take a closer look at enumerative variability, show how it can also be used in the solution space, and briefly discuss why it may be advantageous to do so.

Social Media has become one of the most popular platforms to allow users to communicate, and share their interests without being at the same geographical location. With the rapid growth of Social Media sites such as Facebook, LinkedIn, and Twitter, etc. There is vast amount of usergenerated content. Thus, the improvement in the information quality has become a great Challenge to all social media sites, which allows users to get the desired Content or be linked to the best link relation using improved search / link technique. So introducing semantics to media networks will widen up the representation of the social media networks. Semantic Social Networks representation of social links will be extended by the semantic relationships found in the vocabularies which are known as (tags) in most of social media networks. This paper proposes a new model of semantic social media networks from the perspective of multi-agent systems. The multi-agent system is composed of two main functionalities: semantic indexing and tag ranking.

This poster paper proposes to use a spanning tree communication network to control the formation of a large group of robots. The spanning tree structure minimizes the communication traffic. However, should a loss of connection occur, the formation would break up. This may be solved by dynamic reconfiguration of the communication tree.

Health informatics has become one of the most researched area when it comes to application of Wireless Sensor Networks (WSNs). Wireless and wearable sensors have become prevalent devices to monitor patients at risk for chronic diseases. This helps ascertain that patients comply by the treatment plans and also safeguard them during sudden attacks. The amount of data that is gathered from various sensors are numerous. In this paper we propose to use fog computing to help monitor patients suffering from chronic diseases such that the data is collected and processed in an efficient manner. The main challenge would be to only sort out context sensitive data that is relevant to the health of the patient. Just having a simple sensor-to-cloud architecture is not viable and this is where having a fog computing layer makes a difference. This increases the efficiency of the entire system as it not only reduces the amount of data that is transported back and forth between the cloud and the sensors, but also eliminates the risk that a data center failure bears with it. We also analyse the security and deployment issues of this fog computing layer.

Location prediction is an attractive research area because of its potential application in location-based services such as location-based advertisements, route planning and so on. In this paper, a method based on mining history trajectory logs is proposed to predict people’s next location for a transition from current location. The training samples of this model considers temporal features containing daily and hourly features as well as spatial features like the current location which are all extracted from the historical trajectory dataset. In the learning process, the location-prediction problem can be treated as a multi-class classification and a modified support vector machine (SVM) approach is adopted to learn the classifier models. The method is tested on the Geolife project dataset which is a real life dataset of 182 users over five years. The experimental result shows that the method can provide about a 4% improvement compared with the space time features-based recurrent neural network (STF-RNN) model and a 16% improvement over the Markov models.

This paper explores the formulation of an epistemic approach to Decision Support System (DSS) that have a plurality of various experts affecting said system. In fact, the authors have a DSS computational neuroscience DSS system for neurosurgeons to predict cerebral vasospasms. The problem is that various neurosurgeons will input diametrically opposed indices to the DSS. We propose combining argument theory, with logic to sit on top of our cerebral vasospasm DSS based on rough set theory. We start with integrating the patient’s features into each argument to provide an epistemic approach to an ensemble of decision trees. The goal is to provide neurosurgeons an opportunity to examine their own patient recommendations while countering their DSS responses to the group of neurosurgeons. We propose implementing principles of predicate calculus and formal argument definitions to identify consistency and other measures on the arguments, their associated models and the ontology encompassing them. Once the decision process has been captured, we hope to minimize the decision tree for ease of use.

Since 1998, many e-learning projects initiated by Hong Kong local schools have been granted by the Hong Kong Quality Education Fund (QEF) with the aim of promoting IT in education in Hong Kong. They have generated successful experiences and various quality deliverables. Although all of these valuable good practices are already in place for dissemination to benefit larger audience, there is no easy way to access to them due to the lack of a structural organization amongst different projects. This paper describes some of the good practices extracted from the results of a research framework for evaluating and consolidating good practices from these e-learning projects. We have studied 79 e-learning projects, and suggested 7 good practices that can be reused for e-learning projects. Amongst them, 6 good practices have been compiled as teaching templates as a resource package which is disseminated to local schools.

Recently, with advancement in deep learning technology in CCTV, drone and other various fields, large scale image processing environment becomes crucial and essential for fast real-time processing. In this paper, we shall present a distributed processing environment DeCUVE(Deep Learning Cloud Unified Virtual Environment) which provides scalability and provisioning for the deep learning inference. We shall show that deep learning training inference can be executed fast on our environment.

Machine learning has been applied to various learning problems across a number of disciplines. The availability of data, algorithms and the success of machine learning methods has made machine learning a popular choice for analyzing and solving big data problems. In this paper, we propose a novel machine learning model based on ensemble learning, feature method pairs and multi-view learning concepts. This model targets learning problems with multiple factors or facets. Using botnet detection as an example multi-faceted learning problem, we explain the desirable properties this model promises. The aim of the model is to provide a guideline for designing machine learning based solutions for a specific type of learning problem.

Energy conservation encourages economic growth by redirecting wasteful spending to more productive activities. Energy savings can be used to increase consumption in the broader economy. A more efficient use of energy resources not only conserves this vital resource towards future consumption but also inhibits the excessive exploitation of non-renewable energy resources. The objective of this paper is to develop a suitable framework for an IoT-enabled energy conservation system for an emerging smart community such as Covenant University. A strategic interconnection of sensor networks and energy management systems integrated with the campus IT infrastructure was considered. This energy conservation cyber-physical system is expected to significantly improve energy efficiency in the University and increase energy savings. The scalability of this model will improve energy management in larger communities and cities in Nigeria, a country currently swamped with energy challenges.

Software development companies face the challenge of generating greater value to the business of its customers, understanding and prioritizing their problems, and providing high impact solutions at reasonable costs. This objective requires the use of methodologies that improve productivity through the implementation of new development practices, innovation and motivation of its staff. The application of agile methods adopt a set of principles with the aim of generating profits as adaptation, productivity increase, and therefore improved competitiveness. This paper proposes a model for measuring and improving productivity applicable to Ecuadorian software companies. With this objective, a set of metrics with measurement and interpretation strategies is proposed that provide results in an applicable and practical model which enables progressively improvement of the productivity of the company.

The operation of volunteer computing application depends on the setting of user. Because the number of setting methodologies is diverse, it is difficult for ordinary users to apply an ideal setting to their volunteer applications. In this paper, we propose vcBalancer, a manager which periodically analyze the idle resources of volunteer system and dynamically apply the optimized user setting information to volunteer computing application. Our experiment shows that the vcBalancer improves the overall performance by 3% on average as compared with BOINC client's default configuration.

The human resource performance evaluation process is an important issue for many companies around the world. This paper presents a proposal for a development of a Fuzzy Logic system designed for this goal. Directed to the business environment, the proposed models seeks to accomplish an evaluation with different characteristics directed to basic hierarchical levels of medium and large institutions, considering the need of a different approach to each employee, given the kind of job position that it occupies. The input variables, membership functions and base rules in this article were developed through analysis of similar systems. Two different fuzzy models were developed and applied to the collected data with their results being compared to a decision tree model.

The creative service process is an innovative process that uses newly developed technologies to improve the service models currently used. In this paper, we proposed an IoT-based Cloud-Fog computing platform for the creative service process. The proposed Cloud-Fog computing platform is a distributed computing platform where compute, storage, and network systems of Cloud and Fog are working independently or collaboratively depending on the services performed by the platform. If a service requires resources either from Cloud or Fog, then only the designated Cloud or Fog will be involved. Otherwise, both Cloud and Fog will be collaborated to each other for the service. During the collaboration, the required data may be back and forth in between Cloud and Fog. To demonstrate the usage of the proposed platform for the creative service process, two service examples, smart classroom and city surveillance, are discussed

In this paper, we propose an unsupervised learning-based multi-frame video super-resolution (SR) approach via decision trees model (DTSRV). This novel approach utilizes the temporal redundancy and coherence in consecutive video frames. Motion estimation is applied between consecutive frames to form concatenated motion compensated patches. The low resolution (LR) - high resolution (HR) pairs are then formed to be the training input of the decision trees. After the classification process with decision trees, a linear regression model is learnt to map the relationship between the concatenated LR patches and the HR patches. Results of our experiments show that the approach outperforms state-of-the-art model-based algorithms with an average of 0.97 dB PSNR increase and a much faster speed. It also achieves a 1.4 dB better results for large video sizes than the frame-by-frame image SR using decision trees learning techniques. This is the first time reporting in the literature to use comprehensive random trees/forests structures for video SR. Now the scheme only utilizes two neighbor frames and can already have a good result, which proves its efficiency in real-time application. Our analysis also proves it to have more promising possibilities and advantages for future development.