How to Address Uncertainty in Smaller, Faster, More Agile, Yet Safer Drones?
Erdal Kayacan, Aarhus University, Denmark
Collective and Individual Decision-Making in Swarm Robotics
Sanaz Mostaghim, Computer Science, Otto-von-Guericke-Universität Magdeburg, Germany
Rise of Evolutionary Multi-Objective Optimization: Algorithms and Applications
Kalyanmoy Deb, Michigan State University, United States
Machine Learning with Limited Size Datasets
M. Verleysen, Maching Learning Group, Machine Learning Group, Université Catholique de Louvain, Belgium
How to Address Uncertainty in Smaller, Faster, More Agile, Yet Safer Drones?
Erdal Kayacan
Aarhus University
Denmark
![]()
Brief Bio
Erdal Kayacan received a Ph.D. degree in electrical and electronic engineering at Bogazici University, Istanbul, Turkey in 2011. After finishing his post-doctoral research in KU Leuven at the division of mechatronics, biostatistics and sensors (MeBioS) in 2014, he worked in Nanyang Technological University, Singapore at the School of Mechanical and Aerospace Engineering as an assistant professor for four years. Currently, he is pursuing his research at Aarhus University at the Department of Engineering as an associate professor. He has since published more than 110 peer-refereed book chapters, journal and conference papers in model-based and model-free control, parameter and state estimation, and their robotics applications. He has completed a number of research projects which have focused on the design and development of ground and aerial robotic systems, vision-based control techniques and artificial intelligence. Dr. Kayacan is co-writer of a course book “Fuzzy Neural Networks for Real Time Control Applications, 1st Edition Concepts, Modeling and Algorithms for Fast Learning”. He is a Senior Member of Institute of Electrical and Electronics Engineers (IEEE). Since 1st Jan 2017, he is an Associate Editor of IEEE Transactions on Fuzzy Systems and IEEE Transactions on Mechatronics.
Abstract
Request for increased, almost perfect, accuracy and efficiency of aerial robots pushes the operation to the boundaries of the performance envelope and, thus, induces a need for reliable operation at the very limits of attainable performance. The use of advanced learning algorithms, which can learn the operational dynamics online and adjust the operational parameters accordingly, might be a candidate solution to all the aforementioned problems. This talk will focus both model-based and model-free learning methods to handle various real-time aerial robot control problems. Furthermore, due to the cost associated with data collection and training, the topics related to approaches such as transfer learning will also be mentioned to transfer knowledge between aerial robots and thereby increase the efficiency of their control. Not but not the least, some state-of-the-art drone applications, e.g. autonomous drone racing and fully autonomous cinematography system for aerial drones with the aim of letting the onboard artificial intelligence completely take over the film directing, will also be elaborated.
Collective and Individual Decision-Making in Swarm Robotics
Sanaz Mostaghim
Computer Science, Otto-von-Guericke-Universität Magdeburg
Germany
Brief Bio
Sanaz Mostaghim is a full professor of computer science at the chair of Computational Intelligence and the founder and head of SwarmLab at the Faculty of Computer Science, Otto von Guericke University Magdeburg, Germany. Sanaz holds a PhD degree (2004) in electrical engineering from the University of Paderborn, Germany, has worked as a postdoctoral fellow at ETH Zurich in Switzerland and as a lecturer at Karlsruhe Institute of Technology (KIT), Germany, where she received her habilitation degree in applied computer science. Her research interests are in the area of multi-criteria optimization and decision-making, evolutionary computation, collective learning and decision-making, and their applications in robotics and science. Sanaz is a member of Saxon Academy of Sciences, the vice president of the IEEE Computational Intelligence Society (CIS), IEEE CIS distinguished lecturer, deputy chair of German Informatics and member of several advisory boards. She is an associate editor of IEEE Transaction on Evolutionary Computation as well as member of the editorial board of several international journals on AI. Sanaz has been appointed as a member of advisory board at the Ministry of Infrastructure and Digitalization, State Saxony-Anhalt, Germany.
Abstract
Autonomous systems are becoming more and more ubiquitous and their influence on our lives grows every day. In the last years, computational intelligence methods have – more than ever - extensively contributed to the latest scientific breakthrough in developing such intelligent systems. Nevertheless, one major challenge concerns the real-time reactions of autonomous systems to the unknown dynamics in their environments which is considered to be among the grand challenges in this area. This talk is about multi-objective decision making algorithms in Swarm Robotics. It will give an overview about the design issues and the challenges in real-time applications in robotics and computer games. In most of such applications, the decision makers (robots or agents) must find and select one possible optimal solution in a very limited time frame. This is very challenging, when the environment dynamically changes as the decision maker needs to re-optimize and decide on the fly. Multi-objective decision making algorithms in dynamically changing environments will be addressed and applications in Swarm Robotics will be presented. The results on individual and collective decision making are represented.
Rise of Evolutionary Multi-Objective Optimization: Algorithms and Applications
Kalyanmoy Deb
Michigan State University
United States
Brief Bio
Kalyanmoy Deb is Koenig Endowed Chair Professor at Department of Electrical and Computer Engineering in Michigan State University, USA.Prof. Deb is a pioneer and has been an active proponent of EMO field since 1993. Prof. Deb's research interests are in evolutionary optimization and their application in multi-criterion optimization, modeling, and machine learning. He has been a visiting professor at various universities across the world including IITs in India, Aalto University in Finland, University of Skovde in Sweden, Nanyang Technological University in Singapore. He was awarded IEEE Evolutionary Computation Pioneer Award, Infosys Prize, TWAS Prize in Engineering Sciences, CajAstur Mamdani Prize, Distinguished Alumni Award from IITKharagpur, Edgeworth-Pareto award, Bhatnagar Prize in Engineering Sciences, and Bessel Research award and Humboldt Fellowship fromGermany. He is fellow of IEEE, ASME, and three Indian science and engineering academies. He has published over 549 research papers with Google Scholar citation of over 149,000 with h-index 123. He is in the editorial board on 18 major international journals. More information about his research contribution can be found from https://www.coin-lab.org
Abstract
Started in early nineties, multi-objective optimization problems were solved without any prior preference information. Single-objective evolutionary computation methods were minimally modified to search and store multiple Pareto-optimal solutions simultaneously within an evolving population. The basic idea has not changed in the past three decades, but it has been extended, perfected, and applied to various fields of science, society, engineering, and business. This keynote lecture will present a chronological account of key events and research inventions which propelled the evolutionary multi-objective optimization (EMO) into a field which many novice and expert researchers and applicationists now proudly call it their profession.
Machine Learning with Limited Size Datasets
M. Verleysen
Maching Learning Group, Machine Learning Group, Université Catholique de Louvain
Belgium
![]()
Brief Bio
Michel Verleysen is a Professor of Machine Learning at the UCLouvain, Belgium. He has been an invited professor at EPFL (Switzerland), Université d'Evry Val d'Essonne, Université ParisI-Panthéon-Sorbonne and Université Paris Est (France). He is an Honorary Research Director of the Belgian F.N.R.S. (National Fund for Scientific Research), and the Dean of the Louvain School of Engineering. He is editor-in-chief of the Neural Processing Letters journal (Springer), chairman of the annual ESANN conference (European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning), past associate editor of the IEEE Trans. on Neural Networks journal, and member of the editorial board and program committee of several journals and conferences on neural networks and learning. He is author or co-author of more than 250 scientific papers in international journals and books or communications to conferences with reviewing committee. He is co-author of the "Nonlinear Dimensionality Reduction" book (Springer). His research interests include machine learning, feature selection, nonlinear dimensionality reduction, visualization, high-dimensional data analysis, self-organization, time-series forecasting and biomedical signal processing.
Abstract
Big data are now ubiquitous in many domains of science and technology research, but also in many application areas. Data available in large amounts enable the development of new paradigms for model design, such as deep learning. However there exist countless application contexts where the limited number of data is a concern. An obvious example is patient-based data in healthcare: databases measuring the same information in the same settings for more than a few hundreds or thousands of patients are rare.
Most machine learning methods rely in some way to the approximation of a distribution of data. While such approximation is reasonable when many data are available in a small-dimensional space (small p, large n), it is not in other small data, large-dimensional space contexts (large p, small n); this is the “curse of dimensionality”. Machine learning algorithms may fail in these situations.
This talk will introduce some areas of machine learning that are useful to answer these questions. It will cover fundamental aspects of feature selection, dimensionality reduction, missing data imputation and introduce challenges related to combining data from different sources. Feature selection and dimensionality reduction can lower the dimensionality of the data space, hence enhancing the performances of machine learning methods; missing data imputation and combining data from different sources are seen as ways to take the most of existing data. The talk will be accessible to participants with minimal knowledge of machine learning.