The 25th International Conference on Automation and Computing (ICAC2019)
Lancaster University, UK
5-7 Sep. 2019
Prof. Dong-Ling Xu, Alliance Manchester Business School, The University of Manchester, UK
Bio: Dong-Ling Xu is Chair Professor of Decision Science and Systems at Decision and Cognitive Sciences Research Centre, Alliance Manchester Business School, The University of Manchester, UK. Over the past 30 years, she has been conducting research in data analysis, statistical inference, machine learning, decision support under uncertainty, system development and their applications. She has developed several interactive web and Windows based decision support systems. Those systems are used by organizations such as General Motors, Tesco, NHS, Ford, Shell, BP and CNOOC in a wide range of applications. The ER approach and its software implementation, the IDS software, is used by practitioners and researchers from over 50 countries. She has published over 100 peer reviewed journal papers, book chapters and books.
Title: Hybrid Data-Driven and Knowledge-Based Fraud Detection via Evidential Reasoning
Abstract: In this presentation, a transparent machine learning process is discussed which can help make decisions by acquiring evidence and evaluating the reliability of evidence from both experience and training data. The Evidential Reasoning (ER) rule is used to combine multiple pieces of evidence gathered for multiple feature variables. Each piece of evidence is weighted and combined conjunctively with the weights trained for extended probabilistic inference. Supported by a multinational law firm and UK government, the system is trained by real-world datasets on insurance claims and implemented in to the law firm’s IT system to help monitoring and detecting car insurance fraud for its clients. Results from the application are presented, which reveal that the proposed method has high interpretability and outperforms a number of widely used machine learning models.
Dr Shan Lou, Future Metrology Hub, University of Huddersfield, UK
Bio: Shan Lou is a Senior Research Fellow at the Future Metrology Hub, with the University of Huddersfield.His expertise is surface metrology for additive manufacturing and X-ray computed tomography metrology. He was recently awarded two EPSRC grants to work on surface metrology for additive manufacturing, RCUK Catapult Researchers in Residence fellowship and EPSRC New Investigator grant. Previously, from 2006-2012, he was a R&D engineer with the Hexagon Metrology, a leading global company in the metrology industry. Dr Lou has co-authored 26 refereed journal and 27 conference publications. He is an active participating member of ASTM Committee F42 Additive Manufacturing Technologies, a committee member of the British Standard Institution CT working group, and an editorial board member of the journal ‘Bio-design and manufacturing’. He is a Chartered Engineer of the Institution of Mechanical Engineers and a member of the Institute of Mathematics and its Applications.
Title: Surface Metrology for Additive Manufacturing
Abstract: Additive manufacturing (AM) is paving its way toward the next industrial revolution with wide applications in key industrial sectors, e.g. aerospace, automotive, healthcare, defence. However, many technical barriers still hinder its full commercialisation today. One major issue is that AM processes are not robust enough and AM needs measurement methods for its process control and product verification. This presentation addresses the challenges of measuring and characterising AM surface texture. Various measurement techniques, including tactile, optical and X-ray computed tomography (XCT), are used to measure surfaces produced by selective laser melting (SLM) and electron beam melting (EBM). Their capabilities in measuring AM surface textures are evaluated and compared. Bespoke surface characterisation methods, including feature extraction and parameterisation, are developed in order to reflect the characteristics of AM processes. A case study is given to illustrate how to use XCT to measure an SLM built acetabular implant and how to use the advanced characterisation methods to evaluate the surface quality of the implant in terms of Osseo integration.
Prof. Chenguang Yang, Bristol Robotics Laboratory, University of the West of England, UK
Bio: Chenguang (Charlie) Yang is Professor with Bristol Robotics Laboratory, University of the West of England, UK. He is a Theme Leader of Immersive Teleoperation. He received PhD degree from the National University of Singapore (2010) and performed postdoctoral research at Imperial College London. Prof. Yang has published over 300 international journal and conference papers. He has been awarded EU Marie Curie International Incoming Fellowship, UK EPSRC UKRI Innovation Fellowship, and the Best Paper Award of the IEEE Transactions on Robotics as well as over ten conference Best Paper Awards. His research interest lies in human robot interaction and intelligent system design.
Title: Human-Like Robot Control Design and Human Robot Skill Transfer
Abstract: In the near future, robots are expected to co-habit with our human beings and work closely with us in various fields and even our daily lives. Unfortunately, most of the current robot control technologies are designed for conventional industrial robots which operate behind safeguarding and for predefined tasks, and thus are not able to cope with the varying tasks in unknown dynamic environments. I have therefore developed human-like adaptive control techniques as well as highly effective human robot skill transfer techniques. My work follows the “from human and for human” principle, i.e., study human motor control skills, in order to develop better robot controllers to support human collaborators. My design not only enable versatile and dexterous robot manipulation but also make robot providing personalized assistance to human factors. My investigations not only create a new cross-disciplinary application area where physiologists are able to employ their knowledge and experiences together with roboticists, but will also have a huge impact on the robotics community, through in-depth investigations on the relation between humans and robots.
Short Abstract: My work mainly focuses on the development of new techniques to efficiently transfer human’s adaptive manipulation skills to robots, so that robots can replace people or collaborate with people on the next generation of production lines towards Industry 4.0.
Dr Jing Wang, Department of Computing, Sheffield Hallam University, UK
Bio: Jing Wang gained his BEng degree in Machine and Electronic Technology from the Xidian University, China, in 2006. After graduation, he was appointed as a software engineer and carried out development work on Computer Vision (CV)-based quality control systems. These included assembly line monitoring and industrial robotic controls. In 2008, he became a postgraduate student in the University of Huddersfield and gained his PhD degree at 2012. He then became a research fellow and carried out independent research on image processing, analysing and understanding. Dr Wang joined Sheffield Hallam University in 2017. His interests are focused on the real-world applications of computer vision systems, and he have published more than 20 journal and conference papers in fields related to this. Dr Wang is a member of British Machine Vision Association (BMVA) and British Computer Society (BCS).I also served as chair and editor for International Conference on Automation and Computing.
Title: Machine Learning Tools for Engineering Problems
Abstract: The fast-developing machine learning techniques have been widely used as tools to solve many computer science and engineering problems. This workshop presentation is for our students who wish to start the journey of using the machine learning methods in their researches. I will use a “Hello World” style case study as an example to explain how to develop an end-to-end machine learning system. I hope this topic can let you have an impression of the machine learning system pipeline and have a taste of how to develop a machine learning system from scratch.