Liverpool John Moores University
Paulo Lisboa is Professor in Industrial Mathematics at Liverpool John Moores University and Research Professor at St Helens & Knowsley Teaching Hospitals. He is Fellow of the IMA (Institute of Mathematics & its Applications, UK), Fellow of the IET (Institute of Engineering and Technology, UK) and chair of the Medical Data Analysis Task Force in the Data Mining Technical Committee of the IEEE-CIS. He is on the Advisory Group for Societal Challenge 1: Health, Demographics and Wellbeing in Horizon 2020, the largest coordinated funding programme of health-related research in Europe, which combines medicine and ICT. His research interests are computer-based decision support and data analytics in clinical medicine, public health and sports science, as well as computational marketing. The current focus of interest is on principled approaches to interpret non-linear models. He has over 250 refereed publications with awards for citations and is Associate Editor for IET Science Measure ment and Technology, Neural Computing Applications, Applied Soft Computing and Source Code for Biology and Medicine.
Title: Tell me something I don’t know: the need for interpretation in computational intelligence models
Computational intelligence (CI) models are often evaluated on the basis of predictive performance, lacking appropriate consideration of other aspects than association which might make a claim to the intelligence of the model. Yet appearances can be deceiving, especially with summary performance measures e.g. AUROC. This is especially the case for non-linear models given their ability to exploit any weaknesses in the data, for instance structural artefacts which add a confounding effect over and above the presence of noise. In addition, many applied CI models work well for well classified cases but cannot explain predictions for borderline cases. In other words, they confirm to expert users what they already know but do not add insights to the data in the difficult cases for which CI is most needed. The talk will illustrate some of the pitfalls in the design and validation of databased models. It will then explore principled approaches to interpreting neural networks usin g theoretical methodologies applied to the often opaque maximal separation models driven by computational learning theory and also probabilistic non-linear models from which the geometry of data spaces can be derived. Some important general questions will be explored including the derivation of nomograms for non-linear models, efficiency and interpretability of rule induction, but also a radically different approach to user interfaces for probabilistic classifiers by deriving statistically principled intelligent query systems for case-based reasoning. These models find particular application in clinical medicine where examples will illustrate tumour delineation and detection of response treatment from brain spectroscopy.
Universite' Libre de Bruxelles
Marco Dorigo received his PhD in electronic engineering in 1992 from Politecnico di Milano, Italy, and the title of Agrégé de l’Enseignement Supérieur, from Université Libre de Bruxelles (ULB), in 1995. Since 1996, he has been a tenured Researcher of the fund for scientific research F.R.S.-FNRS of Belgium’s French Community, and a Research Director of IRIDIA, ULB. He is the inventor of the ant colony optimization metaheuristic. His current research interests include swarm intelligence and swarm robotics. He is the Editor-in-Chief of Swarm Intelligence. Dr. Dorigo is a IEEE, AAAI, and ECCAI Fellow. He was awarded the Italian Prize for Artificial Intelligence in 1996, the Marie Curie Excellence Award in 2003, the Dr. A. De Leeuw–Damry–Bourlart award in applied sciences in 2005, the Cajastur International Prize for Soft Computing in 2007, and an ERC Advanced Grant in 2010. In 2015 he will receive the IEEE Frank Rosenblatt Award.
Title: Controlling swarms of cooperating robots
Swarm robotics is about constructing and controlling swarms of autonomous robots that cooperate to perform tasks that go beyond the capabilities of the single robots in the swarm. In this talk, I will give an overview of recent and ongoing research in swarm robotics in my research lab, IRIDIA, at the Université Libre de Bruxelles. In particular, I will present results obtained with homogeneous and heterogeneous swarms of robots that cooperate both physically and logically in search and retrieval tasks.
University of Birmingham
Xin Yao is a Professor of Computer Science at the University of Birmingham, UK. His main research interests include evolutionary computation and ensemble learning. He has had a long-term interest in co-evolution since early 1990s, both for optimisation and learning. He has always been keen on framing co-evolution as an automatic approach to divide-and-conquer in problem solving. His recent work on large scale global optimisation (LSGO) started in 2008, covering more "conventional" evolutionary algorithms as well as estimation of distribution algorithms for either numerical or combinatorial optimisation. Closely related to his practical interest in scaling up evolutionary algorithms, he is also interested in time complexity of evolutionary algorithms, i.e., a more theoretical aspect of scalability.
Title: Large Scale Global Optimisation through Co-operative Co-evolution
Evolutionary optimisation has moved on in recent years from optimising just a few dozens of real-valued variables, although they are still challenging problems. This talk will give a brief overview of some recent efforts towards large scale global optimisation (LSGO) using co-operative co-evolution. Starting the journey from one of the first efforts in optimising problems with up to 1000 real-valued variables , we illustrate new challenges posed by such problems to evolutionary computation approaches and how co-operative co-evolution could be harnessed to address some of those challenges. Then we focus on one of the key issues in LSGO by co-operative co-evolution --- automatic grouping of variables into different co-evolving sub-populations. This is actually a generic and important issue of learning and understanding problem characteristics, especially the interactions among variables. In practice, there is a trade-off to be made between the time we spend on learning problem characteristics and the time we spend on optimisation. Learning makes sense only if the learned information helps to speed up the optimisation more than the time spent on learning. Unfortunately, little is known about the best trade-off. Much work has been based on computational studies, from simple random grouping , which is very fast, to more sophisticated differential grouping , which takes more time in learning. Such grouping methods are not restricted to any particular optimisers used. They can be used in conventional evolutionary algorithms, as well as differential evolution [1,3] and particle swarm optimisation . Similar ideas are applicable to combinatorial optimisation too . This talk will end with a brief discussion of future research directions and how nature inspiration should be considered in problem-solving, e.g., optimisation.
- Z. Yang, K. Tang and X. Yao, "Large scale evolutionary optimization using cooperative coevolution," Information Sciences, 178(15):2985-2999, August 2008.
- M. N. Omidvar, X. Li, Y. Mei and X. Yao, "Cooperative Co-evolution with Differential Grouping for Large Scale Optimization," IEEE Transactions on Evolutionary Computation, 18(3):378-393, June 2014.
- Z. Yang, K. Tang and X. Yao, "Scalability of Generalized Adaptive Differential Evolution for Large-Scale Continuous Optimization," Soft Computing, 15(11):2141-2155, November 2011.
- X. Li and X. Yao, "Cooperatively Coevolving Particle Swarms for Large Scale Optimization," IEEE Transactions on Evolutionary Computation, 16(2):210-224, April 2012.
- Y. Mei, X. Li and X. Yao, "Cooperative Co-evolution with Route Distance Grouping for Large-Scale Capacitated Arc Routing Problems," IEEE Transactions on Evolutionary Computation, 18(3):435-449, June 2014.