There will be five sets of lectures by the distinguished lecturers.
Additional information about abstracts and biographies will be displayed shortly.
Towards the New Science of Big Data Analytics, Based on the Geometry and the Topology of Complex, Hierarchic Systems
These lectures are concerned with pattern recognition, knowledge discovery, computer learning and statistics. I address how geometry and topology can uncover and empower the semantics of data. In these lectures, I lay new foundations for Data Science. These foundations are solidly based on mathematics and computational science. The viewpoint imparted is in line with physicist and Nobel Prize winner, Paul Dirac’s, view that for observed and measured data, as well as virtual and reduced data, mathematics should have primacy even over physics. The hierarchical nature of complex reality is part and parcel of this new, mathematically well-founded way of observing and interacting with (physical, social and all) realities.
1. General Introduction. The Visualization and the Verbalization of Data.
Analytics through the Geometry and Topology of Complex Systems. Metric, Ultrametric Frameworks. Hierarchy and Symmetry.
2. Search and Discovery, Clustering and Regression: Pattern Recognition in Very High Dimensions.
3. Text and Related Analytics. Between Lives of Narratives and Narratives of Lives.
- Social science, following Bourdieu.
- Some issues of cosmology.
- Large data analytics in astronomy, chemistry, finance.
- Literary work, between style and semantics.
- Social media analytics: Letting the data speak.
- Computational psychoanalysis.
Fionn Murtagh is Professor of Data Science in the University of Derby, and also in Goldsmiths University of London. He served for many years as Professor of Computer Science in a number of universities, and for 12 years in the Space Science Department of the European Space Agency. Fionn holds BA, BAI, MSc and PhD degrees in Mathematics, Engineering, Computer Science and Statistics from Trinity College Dublin, Ireland, and Université P&M Curie, Paris 6, France. He has authored or edited 24 books, and published more than 140 journal papers and more than 150 conference papers and book chapters. He is a Fellow of the International Association for Pattern Recognition (IAPR), and a Fellow of the British Computer Society. He is former President of the Classification Society (previously the Classification Society of North America), and former President of the British Classification Society, which celebrated their (joint) 50th anniversary in 2014. He is an elected member of Council, and now the Executive Committee, of the IASC, International Association for Statistical Computing. In regard to scholarly academies, he is a member of the Royal Irish Academy, and of Academia Europaea. Among many editorial responsibilities, he has been Editor-in-Chief the Computer Journal (flagship journal of the British Computer Society, published by Oxford University Press) for more than 10 years. Fionn's research is in data science, digital content analytics and computational science.
Simply Exploring Data
The goal of exploratory data analysis -- or, data mining -- is making sense of data. We develop theory and algorithms that help you understand your data better, with the lofty goal that this helps formulating (better) hypotheses. More in particular, our methods give detailed insight in how data is structured: characterising distributions in easily understandable terms, showing the most informative patterns, associations, correlations, etc. With this as the main theme, I will give three lectures.
In the first lecture we will look into how we can mine *interesting* patterns from data. More in particular, we will see how we can mine compact and non-redundant sets of patterns that describe our data well -- without having to set any parameters, or having to having to make assumptions. We will see that we can do so by taking an Information Theoretic approach, by defining this problem in terms of the Minimum Description Length principle.
In the second lecture we will investigate whether the patterns we so find are also *useful*. We consider a number of exploratory tasks: how to measure and characterise differences between data, how to cluster data without having to choose a distance measure clustering, how to estimate missing values, and how to detect anomalies. All of these are tasks where we want to avoid setting parameters, and where we need insight in why decisions are made.
In the third and final lecture we will up the ante, and try to answer the following questions. Can we derive a causal inference rule to tell whether X causes Y or whether they are merely correlated? Can we do so in a principled manner, without parameters, and without having to assume distributions? Can we do so when X and Y are not iid sets of observations, but objects in general? And can actually instantiate this for multivariate data and efficiently infer causal directions with high accuracy even when the relation is non-linear, complex, non-invertible, and highly non-deterministic? Hopefully not killing any excitement, but so far the answer seems yes.
Jilles Vreeken leads the Independent Research Group on Exploratory Data Analysis at the DFG Cluster of Excellence 'Multimodal Computing and Interaction' at Saarland University in Saarbrücken, Germany. In addition, he is a Senior Researcher in the Databases and Information Systems group of the Max Planck Institute for Informatics.
His research interests include data mining and machine learning, exploratory data analysis and pattern mining. He is particularly interested in developing well-founded theory and efficient methods for extracting informative models and characteristic patterns from large data, and putting these to good use. He has authored over 45 conference and journal papers, 3 book chapters, and has won two best (student) paper awards.
He is program co-chair for ECML PKDD 2016, and was publicity co-chair for IUI 2015, sponsorship co-chair for ECML PKDD 2014, workshop co-chair of IEEE ICDM 2012, and has co-organised seven workshops and four tutorials. He regularly reviews for all the main data mining journals and conferences, and is a member of the editorial board of Data Mining and Knowledge Discovery.
He obtained his M.Sc. in Computer Science from Universiteit Utrecht, the Netherlands. He pursued his Ph.D. at the same university and defended his thesis 'Making Pattern Mining Useful' in 2009 -- for which he received the 2010 ACM SIGKDD Best Dissertation Runner-Up award. Between 2009 and 2013 he was a post-doc at the University of Antwerp, supported by a Fellowship of the Research Foundation -- Flanders.
Robots that Learn: Harnessing Advances in Machine Learning for Smart Actuation
What is your science fiction fantasy: A personal robot butler doing your household chores autonomously or going to the surgeon to buy a new bionic part to augment your body’s capabilities? Today, robots are increasingly making the journey from industry floors to our homes and workplaces – examples include self-driving vehicles (on road and underwater), prosthetic devices, surgical assistants and service robots for drilling, mining and cleaning. Professor Sethu Vijayakumar will explore the scientific challenges in the exciting domain of ‘interactive, autonomous robotics’ and show some of the cutting edge research in topology based representation and planning, variable impedance actuation as well as real time optimal control that is aimed at making robots as versatile, safe, reactive and adaptive as us humans.
The lectures will be structured into three slots. 1. The first will provide an overall picture of key challenges in adaptive, interactive robotics. This intro, along with a live demo of a prosthetic hand, will introduce the concept of how robotics problems can be formulated effectively as some classical problems in machine learning – in domain ranging from humanoid, service to medical robotics. 2. The second lecture will cover supervised machine learning techniques for dynamics and kinematics learning with a special emphasis on non-parametric techniques capable of working incrementally in high dimensions and with real time constraints. The participants will be introduced to and given access to an incremental, locally weighted regression algorithm (LWPR) suitable for use for learning dynamics and kinematics. 3. The third lecture will introduce the participants to optimisation for variable impedance actuation and demonstrate the benefits of employing time varying impedance actuation to dynamic tasks such throwing, hopping and brachiation. The core methodology relies on a formulation of stochastic optimal control and model based plan optimisation. This is particularly relevant to enable safe multi-contact human robot interactions.
Sethu Vijayakumar is the Professor of Robotics in the School of Informatics at the University of Edinburgh, UK and Director of the Institute for Perception, Action and Behavior (IPAB) as well as the co-Director of the Edinburgh Centre for Robotics. Since August 2007, he holds the prestigious Senior Research Fellowship of the Royal Academy of Engineering, co-funded by Microsoft Research. He also holds additional appointments as an Adjunct Faculty of the University of Southern California (USC), Los Angeles and a Visiting Research Scientist at the RIKEN Brain Science Institute, Japan. Prof. Vijayakumar, who has a PhD from the Tokyo Institute of Technology, has pioneered the use of large scale machine learning techniques in the real time control of large degree of freedom anthropomorphic robotic systems including the SARCOS and the HONDA ASIMO humanoid robots, KUKA-DLR robot arm and iLIMB prosthetic hand. His research interest spans a broad interdisciplinary curriculum ranging from statistical machine learning, adaptive control, and actuator design to human motor control and computational neuroscience.
He is the author of over 150 highly cited publications in these fields and the winner of the IEEE Vincent Bendix award, the Japanese Monbusho fellowship, 2013 IEEE Transaction on Robotics Best Paper Award and several other awards from leading conferences. He has been the scientific coordinator and lead PI for a number of national, EU and international research projects, attracting over £25M in research funding over the past 8 years besides serving on numerous EU, DFG and NSF grant review panels and program committees of leading machine learning and robotics conferences. He is a Fellow of the Royal Society of Edinburgh and a keen science communicator with a significant annual outreach agenda.
: 1.28 Informatics Forum, 10 Crichton Street, Edinburgh, EH8 9AB, United Kingdom
AI=Learning to translate
Computational creativity is the study of creative behavior by computational means. This tutorial will introduce students to the central concepts and research questions of the field, with examples of machine creativity in fields such as poetry and music. The emphasis is on conceptual content rather than on specific algorithms. Some creative methods will also be outlined, with emphasis on approaches that learn to create from given data or examples instead of requiring lots of coded expert knowledge within the software or a knowledge base.
Lecture 1. Creativity. Computational creativity. Types of computational creativity. Creativity as the ability to associate in novel ways.
Lecture 2. Creativity as search. Meta-creativity. Machine learning and data mining in computational creativity.
Lecture 3. Social creativity and creative autonomy. Co-creativity. Evaluation of creativity. The FACE model of generative acts.
Hannu Toivonen is Professor of Computer Science at the University of Helsinki, Finland. The current focus of his research is on computational creativity, i.e., on computational processes leading to or supporting creative behavior. His broader research topics include data mining and data science, as well as their applications in natural language engineering, health informatics, bioinformatics, genetics, ecology, and mobile communications. According to Google Scholar, his publications have received 16.000 citations. He also holds 10 patents.
Hannu has chaired the leading conferences in computational creativity (ICCC 2015) and data mining (IEEE ICDM 2014, ICML 2008, ECML 2002, PKDD 2002). He currently serves in the Editorial Boards of Data Mining and Knowledge Discovery and of Machine Learning. Hannu has also been Principal Scientist at Nokia Research Center and visiting scientist at the University of Freiburg, Germany. He is a member of the Finnish Academy of Science and Letters and of the Finnish Academy of Technology.