Elja Arjas (University of Helsinki, Finland)

Professor Elja Arjas is Professor Emeritus at the University of Helsinki, and Research Professor Emeritus at the finnish National Institute for Health and Welfare. He is furthermore a part-time guest researcher at the University of Oslo.

Topics:

Bayesian inference, what and why?

Causal inference: a Bayesian perspective.

Keywords: Inference, time-related aspects (local independence and predictive inference).

Jukka Corander (University of Helsinki, Finland)

Professor Jukka Corander is the group leader for the Bayesian Statistics Group at the University of Helsinki. He earned his PhD degree from Stockholm University in 2000, on the topic of Bayesian learning of graphical models. Scientific areas of particular interest include statistical genetics, bioinformatics, graphical models, stochastic simulation, machine learning, and theory of classification.

Topics:

Population Monte Carlo and population Markov chain Monte Carlo for challenging Bayesian inference problems
Keywords: importance sampling, nonreversible MCMC, adaptive sampling, structural model learning, population algorithms

Approximate Bayesian inference for large intractable models
Keywords: sequential Monte Carlo, dynamic models, non-parametric smoothing, information theory, pseudo-likelihood, ABC

Steve Marron (University of North Carolina at Chapel Hill, United States)

Marron's current interests are in the area of analyzing data that lie in non-standard spaces.  The contexts include High Dimension Low Sample Size (HDLSS) data, and/or data exotic data types, such as manifold and tree-structured data. An overarching framework for this research is Object Oriented Data Analysis (OODA).  This work is motivated by collaborations in cancer research, genetics, image analysis, evolutionary biology, drug discovery and toxicology.  It has spawned a new branch of mathematical statistics: HDLSS asymptotics, where the limiting operation has the dimension growing while the sample size is fixed. Marron's previous theoretical interests were in smoothing methods for curve estimation. These give a flexible and powerful approach to data analysis, especially useful in situations where a good parametric model is unknown, or there is a need for visual model checking. Mathematical analysis, especially a wide array of asymptotics, to the depth of minimax lower bounds, is a frequently used methodological research tool in this area. However computational, numerical and graphical methods are also indispensable. These techniques are broadly applicable in most areas of science where numbers and uncertainty are involved. Personal application areas include biology, economics, geology, human movement, image analysis, marketing, ophthalmology and software engineering.

Topic: Object Oriented Data Analysis

Object Oriented Data Analysis is the statistical analysis of populations of complex objects.  In the special case of Functional Data Analysis, these data objects are curves, where standard Euclidean approaches, such as principal components analysis, have been very successful.  Challenges in modern medical image analysis motivate the statistical analysis of populations of more complex data objects which are elements of mildly non-Euclidean spaces, such as Lie Groups and Symmetric Spaces, or of strongly non-Euclidean spaces, such as spaces of tree-structured data objects.  These new contexts for Object Oriented Data Analysis create several potentially large new interfaces between mathematics and statistics.  The notion of Object Oriented Data Analysis also impacts data analysis, through providing a language for discussion of the many choices needed in many modern complex data analyses.  Even in situations where Euclidean analysis makes sense, there are statistical challenges because of the High Dimension Low Sample Size problem, which motivates a new type of asymptotics leading to non-standard mathematical statistics.

Laura Sangalli (Politecnico di Milano, Italy)

Laura Sangalli is Assistant Professor at  the MOX Laboratory for Modeling and Scientific Computing at Politecnico di Milano.

Topic: Statistical and Numerical techniques for Spatial Functional Data Analysis

Keywords: complex and high-dimensional data, spatial functional data analysis, object oriented data analysis, statistical inverse problems, PDE regularizations, finite elements.

Aila Särkkä (Chalmers, Sweden)

Topic: Challenges in Analysis and Modeling of Spatial Point Patterns

Keywords: anisotropy, cluster process, missing information, nonstationarity, replicated point patterns.

Ernst Wit (University of Groningen, Netherlands)

Professor Ernst Wit is the head of the Statistics and Probability group at the University of Groningen in the Netherlands. He obtained his PhD in Philosophy (1997) on “The Ethics of
Chance: normative reasoning under uncertainty” from the Pennsylvania State University. At PennState he was appointed as a statistics instructor and was the office neighbour of the legendary
C.R. Rao – but no joint lower bound was derived. He spent an academic year at the University of New South Wales in Sydney (1998-1999). In 2000 Wit obtained his PhD in Statistics at the
University of Chicago with Peter McCullagh on the “Categorical Imperative”, an algebraic invariance analysis of statistical models. From 2000 until 2005 he was in the Statistics Department
at the University of Glasgow, where he became a Reader – a lovely English expression for a associate professor. In this period, he got interested in statistical applications in genomics and
published a popular book in 2004 on Statistics for Microarrays. In 2006 he took over the Medical Statistics Unit at the University of Lancaster as Professor of Biometrics. Since 2008 Wit is at the
University of Groningen, where he has continued to work on methodological development in highdimensional inference. Since 1 January 2013 Professor Wit is the European chair of the Bernoulli
Society for Statistics.

Topic: Modelling genomic networks

Networks have become a new paradigm in social, technological and scientific discourse, e.g. social networks, the world wide web, genetic pathways, etc. This development has been accompanied by
new theoretical insights in the mathematical nature of networks. In this course we shall focus on biological networks, which arise in the emerging field of systems biology. The idea is that the functional genome is a stable system, whose emergent properties cannot be described via more traditional gene-by-gene approaches.

The aim of this course is to describe, model and infer biological networks using real genomic data. The course will deal with a large variety of statistical techniques, such as sparse graphical models,
state space models, Boolean networks, hidden Markov models and (stochastic) differential equation models. None of these methods will be assumed to be known. We will consider gene transcription
data (microarrays), but also proteomics and other types of modern high-throughput data.

The course is aimed at statisticians and bioinformaticians with a good knowledge of standard statistical methods (MSc-level Statistics). The applications will be explicitly genetically motivated, but the general outlook of the course is more methodological than applied and can be enjoyed also by a general statistical audience. No explicit biological knowledge will be required.

Laurent Younes (Johns Hopkins University, United States)

Laurent Younes is Professor in Applied Mathematics and Statistics at Johns Hopkins University. Younes' research interests include deformation analysis and computational anatomy, and statistical analysis of image data.

Topic:Statistics in Shape Spaces and Computational Anatomy

An introduction to spaces and manifolds of shapes
Keywords: Riemannian Geometry; Optimal Control; Shape Spaces; Shape Analysis;

Statistical Shape Analysis in Computational Anatomy
Keywords: Statistics on Manifolds; Permutation tests; Medical Imaging; Computational Anatomy; Neuro-degenerative Diseases;