Structural Breaks and Shape Constraints

Home > Workshops > 2022 > Structural Breaks and Shape Constraints

Structural Breaks and Shape Constraints

 16 - 20 May 2022

ICMS, Bayes Centre, Edinburgh

Organisers:

  • Qiyang Han, Rutgers University
  • Axel Munk, University of Goettingen
  • Richard Samworth, University of Cambridge
  • Yi Yu, University of Warwick

About:

Structural break analysis is concerned with the detection and localization of abrupt changes in the data generating distribution in time series and spatial processes. Shape-constrained inference, on the other hand, focuses on automatic learning that adapts to unknown structures of signals. Both topics are well-established in statistics, but the recent explosion of data has resulted in challenges in both fields to find theoretically guaranteed and computationally efficient statistical tools to harness and exploit such structural patterns. These challenges are ubiquitous in many, diverse application areas, such as security monitoring, neuroimaging, financial trading, ecological statistics, climate change, medical condition monitoring, sensor networks, risk assessment for disease outbreaks, flu trend analysis, genetics, electro-physiology and many others.

In the last few years, we have witnessed a growing body of literature in both communities focusing on similar problems, but we are also aware that communication between the two areas could be improved. This workshop focuses on recent developments in structural break analysis and shape-constrained inference, aiming to create a platform to bring the two communities together.

Recorded Talks

Programme:

Please note this programme is subject to change.

Monday 16 May 2022
Registration
Welcome
Jon Wellner, University of Washington Revisiting the symmetric location model: a log-concave perspective
Coffee Break
Carey Priebe, Johns Hopkins University Discovering underlying dynamics in time series of networks
Bodhisattva Sen , Columbia University Multivariate, Heteroscedastic Empirical Bayes via Nonparametric Maximum Likelihood
Lunch
Daren Wang, University of Notre Dame (Online) Optimal High-dimensional Change Point Testing in Regression Settings
Qiyang Han , Rutgers University (Online) Noisy linear inverse problems under convex constraints: Exact risk asymptotics in high dimensions
Coffee Break
Rebecca Willett , University of Chicago (Online) Shape constraints imposed by linear layers in neural networks
Welcome Reception at ICMS
Tuesday 17 May 2022
Claudia Kirch, Otto-von-Guericke University Magdeburg (Online) Data segmentation methodology based on moving sum statistics
George Michailidis, U Florida tbc
Coffee Break
Paul Fearnhead, Lancaster University Fast Online Changepoint Detection via Functional Pruning CUSUM statistics
Zhou Fan , Yale Univeristy Tree-Projected Gradient Descent for Estimating Gradient-Sparse Parameters on Graphs
Lunch
Pierre Bellec, Rutgers University (Online) Data-driven adjustments for confidence intervals and proximal representations in single-index models
Haotian Xu, University of Warwick Change point localisation and inference in high-dimensional regression models under dependence
Coffee Break
Aditya Guntuboyina, University of California, Berkeley (Online) MARS via LASSO
Wednesday 18 May 2022
Lutz Duembgen , University of Bern Isotonic Distributional Regression under Likelihood Ratio Ordering
Min Xu , Rutgers University Root and community inference on preferential attachment networks
Coffee Break
Alexandre Mösching, Universität Göttingen
Oliver Feng, University of Cambridge
Holger Dette, Ruhr-Universität Bochum (Online) Are deviations in a gradually varying mean relevant. A testing approach based on sup-norm estimators
Lunch
Thursday 19 May 2022
Yannick Baraud, University of Luxembourg Robust and adaptive estimation of a density on the line under a shape constraint
Housen Li, University of Goettingen Optimistic search strategy for large scale change point problems
Coffee Break
Cecile Durot, Université Paris Nanterre Unlinked monotone regression
Cun-Hui Zhang , Rutgers University Second- and Higher-Order Anti-Concentration Inequalities, Comparison Theorems and Bootstrap
Lunch
Sabyasachi Chatterjee, University of Illinois at Urbana-Champaign (Online) A Theoretically Tractable Framework for K Fold Cross Vaidation
Kengo Kato , Cornell University (Online) Testing for shape restrictions with U-processes
Coffee Break
Yuting Wei , University of Pennsylvania (Online) Beyond $\log n/\log\log n$ Iterations: Non-asymptotic Analysis for Approximate Message Passing
Dinner at ICMS
Friday 20 May 2022
Tengyao Wang , London School of Economics and Political Science High-dimensional changepoint estimation with heterogeneous missingness
Haeran Cho, University of Bristol High-dimensional time series segmentation under parametric models
Coffee Break
Yining Chen, London School of Economics and Political Science Estimation of S-shaped functions and beyond
Tom Berrett, University of Warwick Optimal nonparametric testing of Missing Completely At Random, and its connections to compatibility
End of workshop, Packed lunch provided