General Information
For Participants

Keynote speakers

Prof. Vijaykrishnan Narayanan
Fellow of IEEE
Pennsylvania State University
Dept. of Computer Science and Engineering
University Park, PA 16802, USA

Keynote Speech Title: Embedded Vision Systems

Abstract: This talk will present challenges and opportunities in the design of embedded vision systems. A confluence of concurrent advances in algorithms, architectures and technology promises new kinds of vision system architectures with cognitive vision capabilities. This talk will discuss the design of such cognitive systems and their use in various applications.

Prof. Minho Jo
Dept. of Computer and Information Science
Korea University
Sejong City 330-700, South Korea

Keynote Speech Title: Selfish Attacks and Detection in Cognitive Radio

Abstract: Cognitive radio is an opportunistic communication technology designed to help unlicensed users utilize the maximum available licensed bandwidth.
Cognitive radio has recently attracted a lot of research interest. However, little research has been done regarding security in cognitive radio, while much more research has been done on spectrum sensing and allocation problems.
A selfish cognitive radio node can occupy all or part of the resources of multiple channels, prohibiting other cognitive radio nodes from accessing these resources. Selfish cognitive radio attacks cause a serious performance degradation problem by not allowing other secondary users to use available licsensed spectrum.
In this keynote speech, he will adress selfish attack types in cognitive radio ad-hoc networks and will introduce selfish cognitive radio attack detection techniques.
In addition, several future research topics in the area will be presented.

Prof. Borko Furht
Professor and Director of NSF Industry/University Cooperative Research Center
For Advanced Knowledge Enablement
Florida Atlantic University, USA

Keynote Speech Title: Industry/University Collaboration in Creating Advanced Multimedia Systems and Tools

Abstract: Integration of data has been the focus of research for many years now. At the metadata level, schema matching and mapping and ontology matching and alignment identify ontological relationships between structured data descriptions (such as attributes, classes, etc.). At the data level, entity resolution (also known as record deduplication) aims at ``cleaning'' a database by identifying tuples that represent the same entity. The need for data integration stems from the heterogeneity of data (arriving from multiple sources), the lack of sufficient semantics to fully understand the meaning of data, and errors that may stem from incorrect data insertion and modifications (e.g., typos and eliminations). With a body of research that spans over multiple decades, data integration has a wealth of formal models of integration, algorithmic solutions for efficient and effective integration, and a body of systems, benchmarks and competitions that allow comparative empirical analysis of integration solutions.
In this talk we will first introduce the NSF-sponsored Industry/University Cooperative Center for Advanced Knowledge Enablement at FAU, which presently has 29 industry members with about $4.5 million memberships. The Center is successfully building a bridge linking academia, industry, and government in a coordinated research initiative. We describe several applied multimedia research projects conducted within the Center including video and image mining for coastline security, 3D image reconstruction and segmentation of brain cells, augmented reality methods for hearing augmentation, automatic asset management in data centers, driver drowsiness detection system using image processing, and a few others. All these projects are initiated by industry partners who are the members of the Center and who are interested to use the obtained research results and create successful commercial products. The talk will complete with our prediction where the multimedia computing is heading in the next several years.

Prof. Jonas Mockus
Institute of Mathematics and Informatics
Vilnius University
Vilnius, LT-08663, Lithuania

Keynote Speech Title: On the New Expert System for Stock Exchange Simulation

Abstract: Stock-exchanges are the traditional source of big data in the form of prices of financial assets. The particular property is that the future prices depend on the investors predictions.
This creates additional difficulties for real life investigation. Thus, one needs an expert system that simulates essential processes of real stock exchange in a controllable manner. The game theory presents the general framework. The talk will be about the new expert system and some surprising, counter-intuitive results which were obtained applying the system to both the historical and virtual data. The virtual data was generated simulating interactions of major investors.
Considerable scientific efforts were, and still is, devoted to the optimization of prediction models and methods. More complex models provide better fitting to historical data, as usual. However, the recent results show that under some, unfavorable, conditions the correlation is negative. To investigate the over-fitting, three time periods representing different economic conditions were tested. To eliminate the influence of temporary and local factors, the virtual stock exchange was developed. This work was designed to investigate both the influence of over-fitting and the relation of profits to the prediction errors. The second problem is more important, since it is almost open area still, in the field of financial optimization. The experiments show that the best investment strategies differ in different time periods and in real and virtual markets. The simpler prediction models provide smaller prediction errors. as usual. This seems natural in the context of over-fitting.

Prof. Nurul I. Sarkar
Auckland University of Technology
Auckland, New Zealand

Keynote Speech Title: Recent Development and Research Activities in Wireless Networks

Abstract: There has been a tremendous growth in the deployment of IEEE 802.11 (a/b/g/n/ac) based wireless local area networks (WLANs) in recent years. This growth is due to the flexibility, low cost, simplicity, and user mobility offered by the technology. Such networks are being deployed widely in homes, offices, schools, shops, hotels, warehouses, factories, and almost anywhere that people live and work. This talk will highlight some of the most recent developments and ongoing research activities in emerging wireless network technologies. Network performance issues and challenges in protocol resign will be discussed. Empirical results will be presented to support discussion. This is a knowledge sharing talk suitable for a general audience.

Prof. Avigdor Gal
Faculty of Industrial Engineering & Management
Technion — Israel Institute of Technology
Technion City, Haifa 32000, Israel

Keynote Speech Title: Big Data Integration

Abstract: Integration of data has been the focus of research for many years now. At the metadata level, schema matching and mapping and ontology matching and alignment identify ontological relationships between structured data descriptions (such as attributes, classes, etc.). At the data level, entity resolution (also known as record deduplication) aims at ``cleaning'' a database by identifying tuples that represent the same entity. The need for data integration stems from the heterogeneity of data (arriving from multiple sources), the lack of sufficient semantics to fully understand the meaning of data, and errors that may stem from incorrect data insertion and modifications (e.g., typos and eliminations). With a body of research that spans over multiple decades, data integration has a wealth of formal models of integration, algorithmic solutions for efficient and effective integration, and a body of systems, benchmarks and competitions that allow comparative empirical analysis of integration solutions.
The evolution of data accumulation, management, analytics, and visualization has recently led to coining the term big data. Big data encompasses technological advancement such as Internet of things (accumulation), cloud computing (management), and data mining (analytics), packaging it all together while providing an exciting arena for new and challenging research agenda. In the light of these landscape changes we analyze in this talk the impact of big data on data integration. In particular, the talk will present advancement in automatic tools for data integration and the changing role of human experts.

Dr. Kevin Deng
Automotive Research Institute
Jilin University
Changchun, China

Keynote Speech Title: Drive Vehicle Intelligence with Virtual Driving

Abstract: Vehicle intelligence has become critically important in coping with the challenges in vehicle fatality and traffic jam faced by automotive industry worldwide. Unlike traditional vehicle development which has been mainly based upon intensive field testing, the development of vehicle intelligence is greatly influenced by its driving environment, which has posed great difficulties due to its large variation, uncertainty and complexity under various driving scenarios and weather conditions.
This talk will present some latest research on tools and method of virtual driving environment aimed to enable high-fidelity design, test and verification of vehicle intelligence. The proposed virtual driving environment includes vehicle models and models of traffic, road and environment sensors such as radar, camera and wireless communication, and a 3D virtual testing field or digital proving ground. Both non real-time and real-time simulation with hardware- and driver- in-the-loop (HIL/DIL) simulation will also be discussed.

Prof. Oleg Burdakov
Division of Optimization
Department of Mathematics
Linkoping University
Linkoping, Sweden

Keynote Speech Title: New network optimization techniques for positioning unmanned aerial vehicles as communication relays for surveillance tasks

Abstract: Many applications for unmanned aerial vehicles (UAVs) are related to surveillance of distant targets, including search and rescue operations, traffic surveillance and forest fire monitoring as well as law enforcement applications. The sensor information gathered must safely be transmitted continuously from a set of surveillance UAVs to a base station via a set of relay UAVs. In many cases, high uninterrupted bandwidth requires line-of-sight between sender and transmitter to minimize quality degradation. Communication range is typically limited, especially when smaller UAVs are used. To accomplish these ends, relay chains for surveillance of a single target, and relay trees for simultaneous surveillance of multiple targets are created. The spatial placement of these UAVs should take into account the local terrain as well as the location of the base station and targets. The surveillance and communication quality of the placement is to be optimised. We show how to reduce these one- and multi-target problems to network optimization problems characterized by a very large number of nodes and arcs. We also present new algorithms developed for effectively solving such problems.

Prof. Amir Hussain
Division of Computing Science and Mathematics
University of Stirling
Stirling FK9 4LA, UK

Keynote Speech Title: Towards Multi-modal Cognitive Control of Next-Generation Autonomous Vehicles

Abstract: Cognitive computation is an emerging discipline linking together neurobiology, cognitive psychology and artificial intelligence. Research in cognitive computation can promote a more comprehensive and unified understanding of diverse topics, including those related to perception, action, attention, learning and memory, decision making, language processing, communication, reasoning, problem solving and consciousness aspects of cognition.
The interdisciplinary field of autonomous vehicle control (AVC) is a rapidly growing one which promises improved performance, fuel economy, autonomy, intelligence, comfort and safety, in next-generation smart cars. One particularly promising alternative to AVC is to break the mission into a set of sub-tasks, each valid over a restricted range of conditions, and to switch between them when required. The problem of selecting from amongst a set of actions or behaviours is also a central problem for animals. There is growing evidence that a set of central brain nuclei - the basal ganglia (BG)- are used by all vertebrates to seamlessly solve this problem. Given the similarity between the problem' domains of AVC and action selection in animals, this challenging research aims to leverage new results from psychology and neurobiology and apply them to AVC.
In this talk, a novel modular cognitive control framework for autonomous systems is presented that could potentially realize the required cognitive action-selection and learning capabilities in our envisaged cognitive machine. In the proposed framework, the BG as a central ‘action-selection’ mechanism in the brain is exploited as the basis of novel BG-based ‘soft-switching’ and ‘dual process’ (automatic and controlled or executive) processing mechanisms. An ongoing case study in autonomous vehicle control is described, as a benchmark problem, with encouraging preliminary results in a range of realistic driving scenarios, such as adaptive cruise control, general path trajectory tracking, lane changes and parallel parking. Current work is focused on extending the cognitive control framework to incorporate dynamic allocation of controllers, on-line learning of salience weights to the BG model, and use of a variable length horizon (as a possible sensorimotor form of chunking) to implement a psychologically motivated dual-process scheme for motion planning and control.
Finally, possible future avenues are explored, including our ongoing work aimed at developing a general modular cognitive framework incorporating multiple modalities, including vision, motor action, language and emotion, required for enabling multi-modal social cognitive and affective behavioral capabilities in future autonomous vehicles.

Prof. Mirjana Ivanovic
University of Novi Sad
Faculty of Sciences
Novi Sad, Serbia

Keynote Speech Title: Modern Machine Learning Techniques and Their Applications

Abstract: During the last several decades machine learning (ML) has become one of the mainstays of information technology and an unavoidable ingredient of contemporary research in a wide range of domains.
One significant view on machine learning is that it represents the modern science of finding patterns and making predictions from large amounts of data, with the goal of creating systems that learn from experience. The field of machine learning borders closely with multivariate statistics, data mining, pattern recognition, and advanced/predictive analytics. Increasing amounts of data available in different areas of human endeavor is good reason to believe that machine learning and other fields concerned with intelligent data analysis will become even more pervasive as a necessary component for technological progress.
The success of machine learning depends on variety of algorithms and techniques that usually belong to one of the main learning styles: supervised, unsupervised, semi-supervised, and reinforcement learning. Relevant techniques include decision tree learning, nearest neighbor methods, partitional/hierarchical/density-based clustering, Bayesian approaches, kernel methods, neural networks and deep learning.
The first part of our presentation will be devoted to a brief overview of most important and frequently used ML techniques and algorithms. We will also outline the common problems learning algorithms face that can hinder performance: the bias-variance tradeoff, under/over-fitting, and high dimensionality of data. In the second part of the presentation we will highlight some of the most recent applications of ML in different areas like: networking, telecommunications, medicine and business.

Prof. Han-Chieh Chao
National Ilan University, Taiwan

Keynote Speech Title: Towards 5G Mobile Communication Technology

Abstract: The third generation (3G) and fourth generation (4G) of mobile phone mobile communication technology have been widely used and launched in the world. Since the infrastructures like Base Stations (BS) and Evolved Node B (eNBs) are deployed everywhere, efficiency is a vital issue for the fifth generation (5G). However, 5G mobile communication is still just being set up. Some technologies and research issues are announced and investigated. This keynote speech introduces the expectative performance goals of 5G. Then, some potential technologies for 5G cellular mobile communications are discussed, such as Millimeter Wave Communication (mmWave), Non-orthogonal Multiple Access (NOMA), Massive MIMO, Cloud Radio Access Network (C-RAN), Heterogeneous Network (HetNet), and so on. Finally, some tentative 5G scenarios are shown and our blueprint for HetNet in 5G is also given.

David G. Stork
Rambus Fellow
Rambus Labs

Keynote Speech Title: Joint design of optics and signal processing: New design principles in computational imaging for the Internet of Things

Abstract: The discipline of computational imaging involves the design of both the optics and the digital signal processing to achieve a desired digital output. The electro-optical system is hence best viewed as an information channel rather than a traditional optical imaging system. Because the intermediate optical image need not “look good,” there is great freedom in the design of the optics—which can be made smaller and cheaper than in “equivalent” systems designed in traditional ways. The burden of digital image creation is thus shifted toward the digital domain, which continues to reduce in cost according to Moore’s Law. Some examples of these new design principles will be presented.

Prof. Yehea Ismail
Director of the Nanoelectronics and Devices Center
American University

Keynote Speech Title: Many-Core Chips: The New High-Performance Computing Platform New design principles in computational imaging for the Internet of Things

Abstract: Sacling as we know it is taking a different direction from the last three decades. Chips with tens of billions of transistors and hundreds of cores are expected to be the future of scaling. These chips will achieve performance through parallelism and application specific optimized cores. This trend will use superior technologies to integrate more cores on a chip rather than to push the frequency envelope as in the past. It is expected that every aspect of design and analysis will need to be modified to accommodate this new platform and trend. There is a clear need for new CAD tools and design methodologies that are very different from existing tools in both their focus and scope. This talk will delve into the specific challenges with respect to both design and CAD that is required for these many core chips. The talk will also provide an overview into the market and technology factors guiding and driving this trend. Attendees will be provided with insight into both present and future research vectors to support this nascent exponential.

Prof. Daoben Li
Beijing Univ. of Posts and Telecomm.
Beijing, China

Keynote Speech Title: A Novel High Spectral Efficiency Waveform Coding –OVFDM

Abstract: Based on the Overlapped Multiplexing Principle, a frequency domain OVFDM (Overlapped Frequency Domain Multiplexing) Coding is proposed. By the data weighted shift overlapped version of any band-limited Multiplexing Transfer Function the coding gain and spectral efficiency are both achieved. The heavier the overlap of the data weighted Multiplexing Transfer Function , the higher the coding gain and spectral efficiency as well as the closer the output to the optimum complex Gaussian distribution. The bit error probability performance is estimated. That shows OVFDM is suitable for high spectral efficiency application and its spectral efficiency is only roughly proportional to SNR.

Prof. Suk-Hwan Suh
Professor and Director of Center for Ubiquitous Manufacturing.
Founding Dean of Graduate School of Engineering Mastership
President-elect of Korean Society of Systems Engineering

Keynote Speech Title: A new paradigm and enabling technologies for product design, manufacturing, and recycling via ubiquitous computing technology

Abstract: Since ubiquitous technology was introduced in the early 1980s, it has rapidly developed, and been applied to various domains mainly for the improvement of human life. In this talk, we propose that ubiquitous computing technology can be effectively used for the design and manufacturing of a product by proposing a new paradigm, called UbiDMR (Design and Manufacture via Ubiquitous Computing Technology). The key aspect of UbiDMR is the utilization of the entire product lifecycle information obtained via ubiquitous computing technology for the design, manufacture, and recycling of the product. The new paradigm can solve many of the problems that have not been properly handled by previous manufacturing paradigms. In the first part, we will address the concept of UbiDMR by the following aspects: 1) why there is a need for UbiDMR, 2) the essence of UbiDMR, 3) enabling technologies, 4) application area, 5) worldwide R&D status, and 6) the societal impacts of UbiDMR. In the second part, enabling technologies for UbiDMR including: 1) Information infrastructure (UPLI: Ubiquitous Product Lifecycle Information), 2) Factory architecture (u-Factory), 3) Product recovery management system (u-PRMS), and 4) Computer Aided Ubiquitous Systems Engineering (CAUSE).
Copyright © CECNet 2014.All Rights Reserved.