Programme - Oral Sessions

Information Technology & Machine Learning
The tremendous success of machine learning and deep learning in complicated problems has been accomplished for last ten years. Starting from the AlphaGo from Google deepmind to notoriously well-known problems in image classification, image recognition, voice regeneration, text generation, and speech recognition, they cannot be modelled and resolved by a classical scientific methodologies and numerical algorithms but smart neural networks with recently developed methodologies such as generative adversary network, variational autoencoder, and recurrent neural network can provide incredibly plausible solutions. It is certainly great success in engineering and application field. However, in machine learning (data-driven modelling ), there are some unsettled factors which cannot be completely understood in a modeling point of view. It is strongly related to fidelity and safety in many applications. In the following sessions, we aim to understand the nature of machine learning, not to only review fancy examples which work very well in applications. Three groups in South Korea which are dedicated to industrial mathematics (industrial & mathematical data analytics research center in Seoul national university, industrial mathematics center on big data in Pusan national university, innovation center for industrial mathematics, national institute for mathematical sciences) will organize the sessions for industrial and applied mathematics in machine learning. The other three sessions are organized by Dr. Kab Seok Kang (Max-Planck Institute for Plasma Physics, Germany), Dr. Young Saeng Park (niversity of Warwick, UK), and PhD. Sogkyun Kim (Rolls-Royce plc, UK). The topics are high performance computing, Advanced Automation for Industry 4.0, and big data in aerospace & automotive engineering.
[ITML1] Recent Progress on Machine Learning and Inverse Problems
Date / Time 2018-08-22 11:00   --   12:40
Room R7
Conveners / Chairs
  • DR. HA, Taeyoung (National Institute for Mathematical Sciences) CONVENER
Machine learning techniques reach to human-level ability in some applications of image and speech recognition recently. The research and development of machine learning algorithms are being expanded in various real-world problems such as communication, security, non-destructive evaluations, medical sciences, that are highly related to nowadays human life. Last decades, various mathematical theories and numerical techniques have been investigated to solve various inverse problems related to medical imaging science and non-destructive evaluations. However, the corresponding problems in many cases are highly nonlinear and ill-posed, with modeling inaccuracies and data uncertainties, which make them difficult to solve and obtain desired results. Machine learning has the potential to deal with these nonlinear and ill-posed problems. It turned out that the accuracy (or decision) of algorithms becomes increasing when one has appropriate training process and enough training data. For this, understanding of modeling inaccuracies, intrinsic data uncertainties, and measurement data process are needed. So, development of an effective machine learning technique to the problem at hand and corresponding mathematical theories must be considered in early stages. In this session, we bring together researchers to present recent results related to inverse problems and deep learning technique in biomedical imaging.
  • PROF. PARK, Won-kwang (Kookmin University) [ 11:00 - 11:40 ]
    Title: Real-time microwave imaging of moving anomaly from scattering matrix
  • DR. HYON, Yunkyong (National Institute for Mathematical Sciences) [ 11:40 - 12:00 ]
    Title: Predicting materials properties of double perovskites using machine learning
  • DR. AHN, Chi Young (National Institute for Mathematical Sciences) [ 12:00 - 12:20 ]
    Title: A mathematical model for estimating the 3D position of left ventricle borders using 2D echocardiography
  • DR. JANG, Jaeseong (NIMS) [ 12:20 - 12:40 ]
    Title: A W-shaped convolutional network for segmentation in echocardiographic images
[ITML2] The Role and Importance of Applied and Industrial Mathematics in the Face of the Fourth Industrial Revolution
Date / Time 2018-08-22 13:30   --   15:10
Room R7
Conveners / Chairs
  • PROF. KIM, Hyun-min (Mathematics, Innovation of Industry, Finance-Fishery-Manufacture Industrial Mathematics Center on Big Data) CONVENER
Applied and industrial mathematics are fields that not only Korean and European government, but almost all countries around the world actively and primarily support to better face the dawn of the 4th industrial revolution. Therefore, it is required to develop an appropriate system to identify, evaluate and diagnose the problems that the affected industries need to solve for the advancement and sustainable growth of world infrastructure. Here though, in order to develop such system, customized mathematical expertise is desperately needed. Nevertheless, as of now it is very rare to utilize math specialists in the current affected industries. In addition, due to the absence of a program that links mathematics with such industries, there are barriers in communication. Therefore, it is time to build a customized applied and industrial mathematics education program and revitalize the fields of applied and industrial mathematics to foster applied and industrial mathematicians with specialized expertise in the world. In order to achieve this, there is an urgent need for active exchanges between experts in applied and industrial mathematics in Europe and Korea. In particular, the exchange with Korean researchers in Europe is at the core. Therefore, in this special session of EKC 2018, we invite the best researchers of applied and industrial mathematics in Europe and they will introduce the recent research trends. Also we will introduce the three industrial mathematics centres in Korea and discuss their roles and further, the general role of applied and industrial mathematics to prepare for the future of the 4th industrial revolution. Finally, we will consider the direction of research on artificial intelligence in Korea and how to promote exchanges with Europe and Korea in the future. Based on this, we want to investigate what kind of desirable roles and talents mathematics, which form the basis of Big Data, Artificial Intelligence, and IoT (Internet of Things) will engage in.
  • PROF. HIGHAM, Desmond (University of Strathclyde) [ 13:30 - 14:10 ]
    Title: Network science from a mathematician's perspective
  • PROF. KIM, Hyun-min (Pusan National University) [ 14:10 - 14:30 ]
    Title: Introduction to Finance•Fishery•Manufacture industrial mathematics center on big data(FFMIMC)
  • DR. LI, Haojun (Yanbian University) [ 14:30 - 14:50 ]
    Title: Applications of adaptive grid method based on multi-resolution analysis to several time-dependent equations
  • MR. KIM, Byungsoo (ETH Zurich) [ 14:50 - 15:10 ]
    Title: Machine learning for fluid simulation in computer graphics
[ITML3 / LSH3] Image detection, analysis, and processing: Innovative Applications and Methodologies
Date / Time 2018-08-22 16:00   --   19:40
Room A1
Conveners / Chairs
  • DR. SONG, Jae Hee (University of Glasgow, UK) COCONVENER
The multi-disciplinary session is a discourse on the current theories, methodologies, and applications of image and signal processing. Specifically, two themes are covered : "Deep Learning for Image Processing : Denoising, Deblurring, and Segmentation” (Theme 1) and "ICT Technologies for Emerging Biomedical Applications"(Theme 2). All who are interested in contributing to either theme are welcome to join. [Theme 1] This session is covered with the mathematical modeling in image processing. Image processing is a method to perform some operations on an image, in order to get an enhanced image or to extract some useful information from it. It is a type of signal processing in which input is an image and output may be image or characteristics associated with that image. Nowadays, image processing is among rapidly growing technologies. It forms core research area within engineering, computer science and applied mathematics. In this session, we will focus on the variational models and numerical methods for various image processing problems. We also investigate deep learning frameworks for image processing. In this session, we focus on the following image processing problems. 1. Image denoising: In many real applications, images unavoidably suffer from noise that occurs during the image-acquisition process. Thus, image denoising is essential in image processing, and it aims to remove noise while simultaneously preserving edges and fine details. 2. Image deblurring: In many applications, images corrupted by a blur and noises due to object movements, hand shaking and out of focus etc. Hence, its goal is to remove blurring and noise while conserving small features and textural patterns. 3. Image segmentation: Image segmentation is a fundamental imaging problem. Its aim is to partition an image into several regions such that the image within each region has uniform characteristics, i.e., edges, intensities, color, and so on. We introduce mathematical modeling based on following approaches. 1. Modeling on variational approaches: To solve image processing problems, we often develop a minimization models based on variational approaches. Variational models have an objective function consisting of a data fidelity term and regularization term. We introduce a novel regularization term or a data fidelity term for each image processing problem. 2. Deep learning: Deep learning is part of machine learning methods based on learning data representations, as opposed to task-specific algorithms. It is very recent method, and gives remarkable results for image processing problems. [Theme 2] ICT and nanotechnology are now extending its contribution to the life sciences and biomedical applications. New concepts and services (e.g., point-of-care, patient-centered care, ubiquitous healthcare monitoring) have been initiated through smart sensors, wearable devices, and novel communication systems. The concept has made (and will make) a significant contribution to improving accuracy, and efficiency in diagnosis and treatment. In this session, experts with various backgrounds are welcomed to share their experience about how they addressed recent issues in biomedical applications.
  • PROF. KANG, Myungjoo (Seoul National University) [ 16:00 - 16:40 ]
    Title: Deep learning and mathematical approach for defect inspection
  • PROF. COCHRAN, Sandy (University of Glasgow) [ 16:40 - 17:20 ]
    Title: Sonopill: the future of capsule ultrasound
  • DR. LEE, Myung-jae (École Polytechnique Fédérale de Lausanne (EPFL)) [ 17:20 - 17:40 ]
    Title: Single-photon detectors for next-generation biomedical applications
  • DR. KANG, Myeongmin (Seoul national university) [ 18:00 - 18:20 ]
    Title: Hybrid total variation based model for image denoising with spatially adaptive parameter selection
  • PROF. KIM, Yunho (Ulsan National Institute of Science and Technology) [ 18:20 - 18:40 ]
    Title: Operator norm minimization for conformational changes in protein structures
  • PROF. JUNG, Miyoun (Hankuk University of Foreign Studies) [ 18:40 - 19:00 ]
    Title: Image segmentation models based on L1 data-fidelity measure
  • MR. JIANG, Hui (Delft University of Technology) [ 19:00 - 19:20 ]
    Title: Energy-efficient bridge-to-digital converters
  • MR. KANG, Eunchul (TU Delft) [ 19:20 - 19:40 ]
    Title: Integrated circuits for 3-D medical ultrasound imaging
[ITML4 / MA4] Latest Advances in High-Performance Computer (HPC) and its Engineering Application
Date / Time 2018-08-23 11:00   --   12:40
Room R7
Conveners / Chairs
  • DR. KANG, Kab Seok (Max-Planck Institute for Plasma Physics, Germany) CONVENER
Computational simulation is nowadays a powerful and indispensable method in analyzing a variety of problems in research, product and process development, and manufacturing. Besides engineering areas, it has been accepted as a powerful methodology even in scientific fields, complementing the traditional approaches of theory and experiment. And also it provides both qualitative and quantitative insights into many phenomena that are too complex to be dealt with by analytical methods or too expensive or dangerous to study by experiments. In this session, we review the current status of High Performance Computer (HPC) and its application to various disciplines. Important aspects of computer science include efficient computing process algorithm and data storage method, distributed networking algorithm, graphics and visualization, and so on. Practical engineering application includes CFD (Computational Fluid Dynamics), FEM (Finite Element Methods), 3D modeling, meshing technique, multi-disciplinary analysis & optimization, uncertainty quantification, statistical analysis, machine learning, etc. several areas.
  • DR. KANG, Ji Hoon (Korea Institute of Science and Technology Information (KISTI)) [ 11:00 - 11:25 ]
    Title: Introduction to KISTI-5 supercomputer and its HPC applications in computational science and engineering
  • PROF. HAN, Woo-suck (Mines Saint-Etienne) [ 11:25 - 11:45 ]
    Title: Construction of morpho-mechanical model of the knee for the prediction of the biomechanical modification induced by a therapeutic action
  • PROF. KIM, Bumsuk (Jeju National University) [ 11:45 - 12:05 ]
    Title: A fully-coupled CFD method for analysis semi-submersible floating offshore wind turbine in wind-wave excitation condition based on the OC5 data
  • DR. KANG, Changwoo (Laboratoire Ondes et Milieux complexes (LOMC), UMR 6294, CNRS-Université du Havre, Normandie Université) [ 12:05 - 12:25 ]
    Title: Large Eddy Simulation of stratified gas-liquid flow in turbulent pipe flow
  • DR. HA, Kwangtae (Fraunhofer IWES) [ 12:25 - 12:40 ]
    Title: Application of numerical analysis with high performance computing on wind turbine for reliability and cost reduction
[ITML5 / MA5] Information Technology Solving the Real World Problems
Date / Time 2018-08-23 13:30   --   15:10
Room R7
Conveners / Chairs
  • DR. KIM, Sogkyun (Nextran Technology Ltd.) CONVENER
A 1958 article in Harvard Business Review referred to Information Technology as consisting of three basic parts: computational data processing, decision support, and business software. Nowadays, information technology is around us everywhere from your mobile phone to Internet network. Data is growing every moment and creating challenges for traditional storage and analytical platforms. In order to efficiently process huge amounts of data into useful business intelligence, information technology requires large amounts of data processing power, sophisticated algorithms, and complicated data analytics. One of the main goals of information technology is to turn raw data into useful and meaningful business intelligence. Therefore, this session is to discuss about the latest development or challenges in information technology to solve the real world problems. This session is open to anyone who is interested in information technology.
  • MR. KIM, Myung-joon (SPRi (Software Policy $ Research Institute)) [ 13:30 - 14:10 ]
    Title: A strategy of software strategy R&D for the fourth industrial revolution
  • DR. NOH, Heeyong (Institute for Manufacturing, the University of Cambridge) [ 14:10 - 14:25 ]
    Title: Development of a taxonomy for promising technology based on text mining analytics
  • MR. KIM, Byungsoo (ETH Zurich) [ 14:25 - 14:40 ]
    Title: Semantic segmentation for line drawing vectorization using neural networks
  • MR. LEE, Wooje (University of Twente) [ 14:40 - 14:55 ]
    Title: Label-free prostate cancer diagnosis using Raman spectroscope and convolutional neural network
  • MR. HWANG, Woochan (Imperial College London) [ 14:55 - 15:10 ]
    Title: Convolutional neural networks for motion artefact reduction in ECGs using accelerometer data
[ITML6] Advanced Automation for Industry 4.0
Date / Time 2018-08-23 16:00   --   17:40
Room R7
Conveners / Chairs
  • DR. PARK, Young Saeng (System Manager, WMG, University of Warwick, UK) CONVENER
Industrial companies are facing strong demand to increase their productivity and reduce their development cost due to the highly globalised market and competition. It is significantly important to adapt enhanced technologies for their success from different perspectives including product design to services for customers. Now, Industry 4.0 has brought not only boost efficiency but also increase effectiveness in order to respond customers’ demands. With the concept from Industry 4.0, it is possible to connect all levels in production and digitally map to all elements of the value chain using cutting-edge technologies such as cyber-physical systems, cloud computing, Internet of Things, machine to machine communication, etc. Companies should examine the benefit of Industry 4.0 and combine it with existing traditional method of production. Advanced automation in particular is one of key principles for Industry 4.0, considering many factors such as smart planning, smart machine, smart data, etc. It demands enhanced software tools to organise sophisticated processes and automate complex operations. It also requires smart machines which have capability to make a link to cloud systems and directly download the approved processing parameters to the machines. Furthermore, the real-time data obtained from production monitoring systems not only from the overall production but also an individual machine will help process-specific decision making and will become increasingly more important to companies. This session aims to look at the current trend and issue of advanced automation for Industry 4.0 as well as discuss on the new technologies with scientists and engineers from Korea and Europe. Participants may have opportunities to find collaboration partners working in the fields. Key words: cyber-physical systems, clouding computing, Internet of things, machine to machine communication, cognitive computing
  • DR. VERA, Daniel (WMG, University of Warwick) [ 16:00 - 16:30 ]
    Title: Digital evolution in manufacturing systems for Industry 4.0
  • DR. AHMAD, Bilal (WMG, University of Warwick) [ 16:30 - 17:00 ]
    Title: Digital solutions for manufacturing industry
  • DR. LEE, Jaejoon (Lancaster University) [ 17:00 - 17:20 ]
    Title: Autonomous feature deployment and composition at runtime for Internet of Things (IoT) applications
  • DR. KO, Youngwook (Queen's University Belfast) [ 17:20 - 17:40 ]
    Title: Cooperative index modulation OFDM with partial relay selection and low-complexity detection designs