A series of lectures in statistics held at different venues in the Netherlands, featuring international and local speakers from the full width of the statistical sciences. Join us to discover new insights and meet new friends!
Check the upcoming seminar, read about the organisation or browse previous editions.
Speakers: Antonio Lijoi (Bocconi University) and Johan Segers (KU Leuven)
Programme:
10:00-11.00 Introductory talk, Antonio Lijoi
11:00-11:15 Coffee break
11:15-12:15 Introductory talk, Johan Segers
12:12-14:00 Lunch break
14:00-15:00 In-depth talk, Antonio Lijoi
15:00-15:15 Coffee break
15:15-16:15 In-depth talk, Johan Segers
16:15- Drinks
Antonio Lijoi
Talk 1: Random discrete structures in Bayesian Nonparametric inference
Discrete random structures, such as random partitions and discrete random measures, play a central role in Bayesian nonparametric modeling and have driven major advances in density estimation, clustering, prediction, feature allocation, and survival analysis. The Dirichlet process (DP) has long served as a reference model, largely due to its analytical tractability. However, its well-known limitations have motivated an active line of research devoted to the development of more general and flexible discrete nonparametric priors. This talk offers an overview of these classes of priors, with particular emphasis on those obtained through the normalization of completely random measures in an exchangeable setting. We discuss characterizations of the resulting random partitions and predictive distributions, and highlight their importance in the design of computational methods for approximate Bayesian inference.
Talk 2: Multivariate species sampling models
Species sampling processes form a cornerstone of Bayesian nonparametrics, underpinning random discrete distributions and exchangeable sequences. When data arise from distinct yet related sources, however, exchangeability is no longer adequate and partial exchangeability provides the appropriate notion of probabilistic invariance. Over the past two decades, numerous dependent nonparametric priors have been proposed in this setting—including hierarchical, nested, and additive constructions: yet a unifying framework has remained elusive.
We address this gap by introducing multivariate species sampling processes, a broad class of nonparametric priors that subsumes most existing models. These processes are characterized by a partially exchangeable partition probability function, which encodes the induced multivariate clustering structure. We establish their main distributional properties and investigate their dependence structure, showing that information sharing across groups is entirely governed by shared ties. This perspective yields new insights into the learning mechanisms of dependent models, including a principled explanation of the correlation structures induced by popular constructions.
Beyond providing a cohesive theoretical framework, our approach offers a constructive basis for developing new models and opens avenues for capturing richer forms of dependence, potentially extending beyond the class of multivariate species sampling processes.
Johan Segers
Talk 1: An introduction to multivariate extreme value theory
Extreme value analysis addresses statistical questions about rare events, such as unusually large observations, extreme risks, or very high quantiles. Typical examples include estimating return levels, assessing the probability of extreme outcomes, or modelling observations that exceed high thresholds. A central challenge in such problems is that the events of interest often lie beyond the range of the available data, making extrapolation unavoidable. Any such extrapolation must therefore be guided by statistical modelling assumptions.
Extreme value theory (EVT) provides a principled framework for this purpose. It identifies limiting models that arise under broad conditions and that constrain the shape of the tails in a mathematically coherent way, in a manner loosely analogous to how the central limit theorem motivates the use of the Gaussian distribution. The resulting models are well suited for extrapolation, but differ substantially from classical parametric families.
The lecture offers an introduction to multivariate extreme value theory, with an emphasis on how extreme outcomes in different variables tend to occur together. Adopting a copula-like separation of marginal behavior and dependence, the focus will be on the latter. Several complementary viewpoints on multivariate extremes will be outlined, each providing intuition for the structure of extremal dependence. The lecture will conclude with a brief discussion of a small number of model families commonly used in practice.
Talk 2: X-vine models for multivariate extremes
Based on joint work with Anna Kiriliouk and Jeongjin Lee.
Regular vine sequences permit the organization of variables in a random vector along a sequence of trees. Regular vine models have become greatly popular in dependence modelling as a way to combine arbitrary bivariate copulas into higher-dimensional ones, offering flexibility, parsimony, and tractability. In this project, we use regular vine structures to decompose and construct the exponent measure density of a multivariate extreme value distribution, or, equivalently, the tail copula density. Although these densities pose theoretical challenges due to their infinite mass, their homogeneity property offers simplifications. The theory sheds new light on existing parametric families and facilitates the construction of new ones, called X-vines. Computations proceed via recursive formulas in terms of bivariate model components. We develop simulation algorithms for X-vine multivariate Pareto distributions as well as methods for parameter estimation and model selection on the basis of threshold exceedances. The methods are illustrated by Monte Carlo experiments and a case study on US flight delay data.
The Van Dantzig seminar is a nationwide series of lectures in statistics, which features renowned international and local speakers, from the full width of the statistical sciences. The name honours David van Dantzig (1900-1959), who was the first modern statistician in the Netherlands, and professor in the “Theory of Collective Phenomena” (i.e. statistics) in Amsterdam. The seminar will convene 4 to 6 times a year at varying locations, and is supported financially by among others the STAR cluster and the Section Mathematical Statistics of the VVSOR.
If you have any comments, questions, requests, suggestions etc. please contact the organizers.
Eni Musta (University of Amsterdam)
Hanne Kekkonen (TU Delft)
Mathisca de Gunst (VU Amsterdam)
Geurt Jongbloed (TU Delft)
Aad van der Vaart (TU Delft)
Biography of David van Dantzig from MacTutor History of Mathematics archive
David van Dantzig’s statistical work by J. Hemelrijk
The Statistical Work of David Van Dantzig (1900-1959) by J. Hemelrijk
24 October 2025 (Vrije Universiteit Amsterdam)
Gerda Claeskens (KU Leuven)
Talk 1: Obtaining valid inference after variable selection
Talk 2: Selective inference in graphical models after edge selection
Andrew Duncan (Imperial College London):
Talk 1: Learning to Sample: Bridging Classical Monte Carlo and Modern Generative Models
Talk 2: Score-based Methods for Generation and Sampling
21 March 2025 (Vrije Universiteit Amsterdam)
Davy Paindaveine (Université Libre de Bruxelles): Rank tests for PCA under weak identifiability
Elisa Perrone (TU Eindhoven): Measuring association of zero-inflated data
31 January 2025 (TU Delft)
Tapio Helin (LUT University): Next frontier of Bayesian Inverse Problems: Optimal Experimental Design
Mark van der Wilk (University of Oxford): Bayesian Model Selection from Gaussian Processes to Deep Neural Networks
22 November 2024 (TU Delft)
Angelika Rohde (Universität Freiburg): Nonparametric Bootstrap of High-Dimensional Sample Covariance Matrices
Thomas Verdebout (Université Libre de Bruxelles): Asymptotic power of Sobolev tests for uniformity on hyperspheres
4 October 2024 (Vrije Universiteit Amsterdam)
Veronika Rockova (Booth School of Business at the University of Chicago): Adaptive Bayesian Predictive Inference in High-dimensional Regression
Paul Doukhan (Cergy-Paris Université): Weak dependence, properties and some applications
Friday, June 7 2024 (TU Delft)
Ismaël Castillo (LPSM, Sorbonne Université): A tale of heavy tails
Jana de Wiljes (TU Ilmenau): Data assimilation: theory, algorithms and applications
April 12, 2024 (Vrije Universiteit Amsterdam)
Sara van de Geer (ETH Zurich): Rates of convergence for tensor denoising
Michael Sørensen (University of Copenhagen): Models of time series of angular data: diffusion processes on the torus
Ernst Wit (Universita della Svizzera italiana): From Toegepaste Statistiek to Causality and back
February 2, 2024
Yannick Baraud (University of Luxembourg): From robust tests to robust Bayes-like posterior distributions
Siem Jan Koopman (VU Amsterdam): Nonlinear non-Gaussian state space models
University of Amsterdam
December 1, 2023
Sofia Olhede (EPFL) – On Graph Limits as Models for Interaction Data
Björn Sprungk (TU Bergakademie Freiberg) – Noise-level robust sampling and Bayesian inference on the sphere
Delft University of Technology
September 29, 2023
Mark Podolskij (University of Luxembourg)
Saskia le Cessie (Department of Clinical Epidemiology and Department of Biomedical Data Sciences, Leiden University Medical Center)
VU Amsterdam
June 2, 2023
Francois Carron (Oxford),
Christian Robert (Paris-Dauphine)
CWI
April 14, 2023
Bärbel Finkenstädt (Warwick)
Randolf Altmeyer (Cambridge)
Delft University of Technology
February 10, 2023
Judith Rousseau (Oxford)
Chris Sherlock (Lancaster)
VU Amsterdam
December 16, 2022
Speakers: Aretha Teckentrub, Christophe Giraud
University of Amsterdam
June 17, 2022
Speakers: Rajen Shah, Alexis Derumigny
VU Amsterdam
October 22, 2021
Speakers: Sophie Langer, Robert Scheichl
Delft University of Technology
April 30, 2021
Speakers: Ryan Martin, Aaron Smith
Online
October 16, 2020
Speakers: Fadoua Balabdaoui, Yoav Zemel
Online
February 14, 2020
Speakers: Domenico Marinucci, Thomas Nagler
Delft University of Technology
November 29, 2019
Speakers: Gareth Roberts, Holger Dette
VU Amsterdam
October 18, 2019
Speakers: Julio Backhoff, Nestor Parolya
Leiden University
May 24, 2019
Speakers: Louis Aslett, Adrien Saumard
Delft University of Technology
April 5, 2019
Speakers: Lorenzo Rosasco, Chris Oates, Paul Fearnhead
VU Amsterdam
15 February 2019
Speakers: Carola-Bibiane Schönlieb, Antonietta Mira
University of Amsterdam
7 December 2018
Speakers: Christoph Brune, Olga Klopp
Leiden University
26 October 2018
Speakers: Gerard Kerkyacharian, Stefan Sommer
Delft University of Technology
28 September 2018
Speakers: Rui Castro , Estate Khmaladze
VU Amsterdam
25 May 2018
Speakers: Ed George , Barry Schouten
University of Amsterdam
9 March 2018
Speakers: David Dunson , Quentin Berthet
Leiden University
26 January 2018
Speakers: Geert Molenberghs, Alexandre Tsybakov
Delft University of Technology
15 December 2017
Speakers: Petros Dellaportas, Catherine Matias
VU Amsterdam
13 October 2017
Speakers: Andrew Parnell, Alberto Roverato
University of Amsterdam
2 June 2017
Speakers: Elisabeth Gassiat, Johanna Ziegel
Leiden University
6 April 2017
Speakers: Tatyana Krivobokova, Botond Szabó
Delft University of Technology
27 January 2017
Speakers: Cun-Hui Zhang, Eric Moulines
University of Amsterdam
2 December 2016
Speakers: Alexandra Carpentier, Alois Kneip
VU Amsterdam
26 October 2016
Speakers: Jim Griffin, Jakob Söhl
Leiden University
24 June 2016
Speakers: Axel Munk, Gilles Blanchard
University of Amsterdam
1 April 2016
Speakers: Omiros Papaspiliopoulos, Cristina Butucea
Delft University of Technology
26 February 2016
Speakers: Iain Johnstone, Jelle Goeman, Kolyan Ray
VU Amsterdam
8 October 2015
Speakers: Martin Wainwright, Wolfgang Polonik, Giulia Cereda
Leiden University
16 April 2015
Speakers: Sara van de Geer, Gernot Müller
Leiden University
6 March 2015
Speakers: Marc Hoffmann, Ivan Vujacic, Moritz Schauer
University of Amsterdam
27 November 2014
Speakers: Lutz Dümbgen, Jesper Møller
Delft University of Technology
9 October 2014
Speakers: Arnak Dalalyan, Marco Grzegorczyk
VU University Amsterdam
11 April 2014
Speakers: Johan Segers, Richard Gill, Tina Nane
Leiden University
28 February 2014
Speakers: Andrew Gelman, Harrison Zhou
University of Amsterdam
31 January 2014
Speakers: Richard A. Davis, Nicolai Meinshausen
Delft University of Technology
12 December 2013
Speakers: Pascal Massart, Joris Mooij
VU University Amsterdam
12 September 2013
Speakers: Jon Wellner, Paulo Serra
Leiden University
3 June 2013
Speakers: Richard Samworth, Subhashis Ghosal
University of Amsterdam
