Special Sessions

Copulas

Copulas have become a central tool in modern multivariate statistics, offering a flexible framework to model and analyze complex dependence structures beyond the limitations of classical correlation. By separating the marginal behavior of random variables from their dependence structure, copulas allow researchers to capture non-linear, asymmetric, and tail dependencies that often occur in real-world data. The session will cover both theoretical advances and practical applications of copula models. Topics include novel approaches to copula construction and estimation, dependence measures, inference procedures, and goodness-of-fit testing. Special emphasis will be placed on recent developments in high-dimensional settings, time-series contexts, and dynamic copula models.

Organized by: Ostap Okhrin (Germany)

Projection Pursuit

Projection pursuit is a multivariate statistical technique aimed at detecting interesting low-dimensional data projections. It looks for the data projection which maximizes the projection pursuit index, that is a measure of its interestingness. After an interesting projection is found, it is removed to facilitate the search for other interesting features. Projection pursuit addresses three major challenges of multivariate analysis: the curse of dimensionality, the presence of irrelevant features and the limitations of visual perception. Its applications have been hampered by computational, interpretative and inferential problems. Additional problems arise when data are high-dimensional, that is when there are more variables than units. This session outlines the main features of projection pursuit and its connections with other multivariate techniques. The theory is illustrated with both real and simulated datasets.

Organized by: Nicola Loperfido (Italy)

New Trends in Robust Statistical Analysis - theory and application

Recent research clearly shows that classic statistical approaches (e.g., those related to model parameter estimation) may be insufficient for describing phenomena observed in reality. There are many reasons for this, for example, the fact that in industrial data we observe impulsive behaviors, which indicate that the models describing the data are non-Gaussian, or in biological experiments (so-called single particle tracking), where we observe behaviors corresponding to models with random parameters. During this session, new statistical approaches and methods used to describe and analyze data with a complex structure will be presented. The theoretical results presented will be supported by real-world examples of data analysis from various fields.

Organized by: Agnieszka Wyłomańska (Poland)

Modeling of Covariance Structure for Continuous and Discrete Dependent Data

This session includes several talks on modeling of covariance structure for continuous and/or discrete dependent data. Firstly, deep leaning method is applied to modelling of covariance structure for longitudinal data and its advantage over the existing linear or nonlinear model assumption-based methods are presented. Secondly, modeling of covariance structure for spatially discrete data, particularly spatially binary data, is provided through using Copula approach. Thirdly, modeling of covariance structure of mixed longitudinal data, including continuous, binary and ordered longitudinal data, is addressed. The proposed methods are applied to real data examples arising in practices.

Organized by: Jianxin Pan (China)

Inference for Non-Gaussian Distributions in Linear Models

Non-Gaussian features in data can be seen either as a curse or a blessing, depending on one's perspective. For those who value the elegance of the Gaussian paradigm and its well-established theoretical results, the disruptive effects of non-Gaussianity can be a source of frustration, challenging long-standing frameworks. Yet for others, these same departures from normality offer opportunities: by accounting for non-Gaussian structures, one can develop more efficient inference procedures and more accurate predictive methods, moving beyond the familiar—if somewhat predictable—Gaussian path. This session explores the world beyond the Gaussian paradigm by presenting new theoretical foundations and innovative inferential methodologies designed to address non-Gaussian data.

Organized by: Krzysztof Podgórski (Sweden)

Design and Analysis of Experiments

Organized by: Radoslav Harman (Slovakia)

Machine Learning and Statistical Inference

The session focuses on current trends at the intersection of machine learning and classical statistical inference. It covers topics such as model interpretability, uncertainty quantification, and extensions of linear methods for analyzing large and complex data sets. The aim is to highlight how statistical principles contribute to the development of reliable and transparent AI models.

Organized by: Tomasz Górecki (Poland)