Three days of Statistical learning

Statistical learning, more commonly known as data mining, is traced back along three family lines: The longest of these three lines is classical statistics. Without statistics, there would be no data mining, as statistics are the foundation of most technologies on which data mining is built.

The second longest family line is artificial intelligence, or AI. This discipline, built upon heuristics as opposed to statistics, attempts to apply human-thought-like processing to statistical problems. The promoters of this line was mainly psychologists, neurobiologists and later computer scientists.

The third family line is machine learning, which is more accurately described as the marriage of statistics and AI. Machine learning attempts to let computer programs learn about the data they study, such that programs make different decisions based on the qualities of the studied data, using statistics for fundamental concepts, and adding more advanced AI heuristics and algorithms to achieve its goals.

Why Statistical learning?

The discipline statistical learning has grown out from so diverse fields as psychology, neurobiology, mathematics, statistics, image analysis, medicine, business intelligence, robotics just to mention a few. This has created an overgrown flora of different terms for one and the same subject which make it hard to grasp the subject.

This summer school is the second one to spread knowledge about statistical learning but it is the first with the objective to bring together statisticians from both the industry and management (public and private) not forgetting researchers and students at universities and colleges.

Our lecturers

John Shawe-Taylor

is a professor at University College London where he is Director of the Centre for Computational Statistics and Machine Learning . His main research area is Statistical Learning Theory, but his contributions range from Neural Networks, to Machine Learning, to Graph Theory.

John's aim is to introduce Statistical Learning Theory: aims, some key techniques and classical results, and time permitting more recent methods (eg Rademacher complexity, PAC-Bayes analysis). He is currently the scientific coordinator of a Framework VI Network of Excellence in Pattern Analysis, Statistical Modelling and Computational Learning .

Mattias Villani

is an Associate Professor at Stockholm University and an Adviser at Sveriges Riksbank. His main research areas are Bayesian Statistics, Statistical Learning and Econometrics.

Mattias' aim is introduce Bayesian analysis and computations followed by a selective review of some flexible Bayesian models that have been found useful in statistical learning applications. Examples of such models are nonparametric regression models for discrete and continuous response variables, copula models for multivariate data, finite mixture models and Dirichlet process mixture models, and graphical models.

Jennifer Castle

is an Economics Fellow at Oxford University and a member of the Institute for Economic Modelling, Oxford Martin School, University of Oxford. Her research interests lie in the fields of model selection and forecasting, focusing on general-to-specific model selection.

Jennifer’s aim is to introduce the theory of reduction, which provides the basis upon which general-to-specific model selection is conducted. Recent advances in selection, including bias correction, impulse-indicator saturation to identify outliers and breaks, more variables than observations, nonlinear models, perfectly collinear combinations, and implementing theory restrictions will be discussed.