Accepted Papers
  • Accelerated Bayesian Optimization For Deep Learning
    Ayahiko Niimi and Kousuke Sakamoto, Future University Hakodate,Japan
    ABSTRACT

    Bayesian optimization for deep learning has extensive execution time because it involves several calculations and parameters. To solve this problem, this study aims at accelerating the execution time by focusing on the output of the activation function that is strongly related to accuracy. We developed a technique to accelerate the execution time by stopping the learning model so that the activation function of the first and second layers would become zero. Two experiments were conducted to confirm the effectiveness of the proposed method. First, we implemented the proposed technique and compared its execution time with that of Bayesian optimization. We successfully accelerated the execution time of Bayesian optimization for deep learning. Second, we attempted to apply the proposed method for credit card transaction data. From these experiments, it was confirmed that the purpose of our study was achieved. In particular, we concluded that the proposed method can accelerate the execution time when deep learning is applied to an extremely large amount of data.

    Data Center Network (DCN) Architecture
    Eyob Semere Ghebremicael,University of Stuttgart, Germany
    ABSTRACT

    Nowadays, Data Centre Network (DCN) architectures consist thousands of resources such as servers, routers, switches etc. Data Centers are growing dynamically and the fundamental challenge is on how to interconnect the devices within it. The growth of Data centers implies to these physically connected devices as well as to the robust application services it supports. Efficient data center network topologies need to handle the demands requested as a result of the growth. Data center network topologies have to address the basic design issues. The traditional network topology isnít cost-effective to handle the demands of the growing data centers. Even though it deploys expensive devices, it couldnít satisfy the needs of data centers. Recently some network topologies that deploy commodity devices are introduced as an alternative solution. This paper focuses on three approaches of topologies that deploy commodity devices; Fat Tree, Dcell and Jellyfish. It analyses and compares these topologies and against the traditional one.

    Scheduling job seekers using Genetic Algorithms
    Hussain M. Alfadhli,Kuwait
    ABSTRACT

    This paper applies the heuristic algorithms to the problem of scheduling the job seekers over the available jobs in the public sector (government). As a way of providing equal chances for the graduates to win a job in government sector, the government of Kuwait uses a central scheduling system to nominate the graduates of the colleges and universities over the available job vacancies in the government. Usually the government uses a very classical scheduling system to nominate job seekers to the available jobs. The current scheduling method uses very rigid criteria which considers some parameters like Major, GPA, Graduation year, Marital status in a sequential order to evaluate the job seekers. We will use the genetic algorithms to get an optimal scheduling to job seekers over the available jobs. The aim of this project is to make the process of assigning the person for the future job as perfect and fair as possible. We will put in mind two main criteria. The first is to put the right person (in term of educational degree) in the right position (the job duties) depending on the list of approved certificates to hold the job for each job title. The second criterion is to put the person in a job he likes depending on a list of desired jobs selected by the person himself.

    The Five Layers of the Internet on the Computing Level
    Bing Li, XiíAn Technological University,china
    ABSTRACT

    To share huge amount of heterogeneous information, the Internet is reconstructed to consist of five layers, including routing, multicasting, persisting, presenting and humans.Routing layer establishes the fundamental substrate and locates resources with social disciplines. Multicasting layer disseminates data efficiently based on the routing. Persisting layer accesses persistent data with minimum dedicated resources. Presenting layer absorbs usersí interactions to adjust the underlying layers through connected views to users. Different from the lower layers, the topmost one is made up with humans, which are social capital dominating the Internet. Within the upgraded infrastructure, besides the situation that a lower layer supports its immediate upper one, the humans layer influences the lower ones by transferring social resources to them. Those resources lead to adaptations and adjustments of the software layers since each of them needs to follow social rules. Eventually, the updated underlying layers return latest outcome to users upon those modifications.

    Adapting Naive Bayesian Classifiers in Adaptive MOOCs based on Intended Learning Outcomes
    Duaa Abu Samra1 and Ahmed Ewais2, 1Arab American University,Jenin PALESTINE, 2Vrije Universiteit Brussel,Brussels BELGIUM
    ABSTRACT

    The basic idea of this paper is designing and implementing an adaptive MOOCs system that maps the intended learning outcomes ILOs to learning concepts and then to learning resources for specific subject, by adapting NaÔve Bayesian Classifier techniques. This system allows the learner to select one or more ILOs that he wants to study it. Accordingly to these ILOs, NaÔve Bayesian provides the required learning resources from the same course which are called related learning resources, and from other courses related to the same subject which are considered as recommended learning resources. Taking into account the learner style that are usually defined by learner himself and stored in learner profile. Furthermore, it presents the learning path that is identified by the pedagogical relationship between learning concepts for each intended learning outcome. So that, the learner must follow it to complete all selected intended learning outcomes. Finally, we evaluates the system in two phases: first by calculating the precision and recall between the ILOs and the learning resources. Second, testing the system by using it from group of learners, then analyzing their answers on a set of questions that were developed to measure the efficiency of the system in terms of its ability to satisfy the learner and enable him to complete the intended learning outcomes that he wants to learn.

    Cross-channel Interference Mitigation Technique in 5G LTE Network by Resource Scheduling
    Shahadate Rezvy1 and Tahmina Zebin2, 1Middlesex University ,2University of Manchester, UK
    ABSTRACT

    5G Cross-channel interference which is occurred by using same channel in Macro-Femto cellular network is one of the constraints in high-speed data networks such as LTE. These issues are particularly challenging to deal in networks that serve real time flows. The main focus of this paper is to develop a cooperative scheme that exploits the total available bandwidth resources and mitigate the cross-channel interference through improving user throughput between deployed Femtocells and Macrocells in 5G LTE cellular network. As a part of this focus, this paper presents a novel resource scheduling algorithm technique to evaluate the performance of the proposed algorithms in the context of crosschannel interference mitigation.