Big Data Inverse Problems

Scientific organisers

Matthias Chung, Emory University
Matthias Ehrhardt, University of Bath
Carola Bibiane Schönlieb, University of Cambridge

About:

The rapidly evolving field of data science recognizes the urgent need for novel computational methods to overcome challenges of inference and uncertainty quantification to make informed decisions in big data settings. Emerging fields such as data analytics, machine learning, and uncertainty quantification rely heavily on efficient computational methods for inverse problems.

This workshop will gather researchers from the intersecting fields of inverse problems, uncertainty quantification, data analytics, machine learning, and related areas to discuss novel theory and new methods for big data challenges. A focused multifaceted research group gathered at one location is an excellent opportunity for established and early career researchers. The aim of this workshop is to foster a diverse, open, and friendly environment to advance this field, open new directions, and to stimulate new collaboration.

Due to a recent uptick in the activity of registration and accomodation scammers, details of who is speaking is available to attending delegates via request.

Programme:

	MONDAY 20 MAY 2024
10:00-10:20	Registration and Refreshments
10:20-10:30	Welcome and Housekeeping
10:30-11:00	Per Christian Hansen, DTU	AB- and BA-GMRES Methods for X-Ray CT with an Unmatched Back Projector	Abstract
11:00-11:30	Marcelo Peyrera, Heriot-Watt University	Equivariant Bootstrapping for Uncertainty Quantification in Imaging Inverse Problems	Abstract
11:30-12:00	Jonas Latz, University of Manchester	Losing momentum in continuous-time stochastic optimisation	Abstract
12:00-14:00	Lunch
14:00-14:30	Anupam Gumber, Machine Learning Genoa Centre, University of Genoa	Generative Fourier neural operators for non linear inverse problems	Abstract
14:30-15:00	Tia Chung, Emory University	Big Data Inverse Problems — Promoting Sparsity and Learning to Regularize	Abstract
15:00-15:30	Refreshments
15:30-16:00	Tobias Wolf, University of Klagenfurt	Nested Bregman Iterations for image decomposition	Abstract
16:00-16:30	Luca Calatroni, CNRS	Bregman relaxation of $\ell_0$-regularized criteria with general data terms	Abstract
17:00-18:00	Welcome Reception, hosted at ICMS
	TUESDAY 21 MAY 2024
9:15-9:45	Simon Arridge, UCL	Learned image flows with equivariance constraints	Abstract
9:45-10:15	Ferdia Sherry, University of Cambridge	Designing Stable Neural Networks using Convex Analysis and ODEs	Abstract
10:15-10:45	Refreshments
10:45-11:15	Francesco Tudisco, University of Edinburgh
11:15-11:45	Mohammad Sadegh Salehi, University of Bath	An adaptively inexact first-order method for bilevel learning	Abstract
11:45-12:15	Carola-Bibiane Schönlieb, University of Cambridge	Machine learned regularisation for inverse problems - the dos and don’ts
12:15-14:15	Lunch
14:15-14:45	Yves Wiaux, Heriot-Watt University	R2D2: a deep neural network series for ultra-fast large-scale imaging in radio astronomy (but not only)	Abstract
14:55-15:15	Hong Ye Tan , University of Cambridge	Data-driven geometry for convex optimization	Abstract
15:15-15:45	Refreshments
15:45-16:15	Elizabeth Newman, Emory University	Optimal Tensor Algebras for Efficient Data Representation	Abstract
16:15-17:00	Young Researcher Gathering; Carola + Panel discussion
18:00-19:30	Samuli Siltanen, Public Lecture, hosted in G.03 (ground floor), University of Helsinki	The Magic of Maths: 3D X-ray vision
	WEDNESDAY 22 MAY 2024
9:15-9:45	Omar Ghattas, University of Texas at Austin	Derivative-informed neural operators (DINOs) for Bayesian inverse problems governed by PDEs	Abstract
9:45-10:15	Patrick Fahy, University of Bath	Learning preconditioners for inverse problems	Abstract
10:15-10:45	Refreshments
10:45-11:15	Ronny Ramlau, Johannes Kepler University Linz	Reconstructability issues for the atmospheric tomography problem in adaptive optics for earth bound telescopes
11:15-11:45	Markus Rau, Newcastle University
11:45-12:15	Silvia Gazzola, University of Bath	Automatic space-variant anisotropic Tikhonov regularization through bilevel optimization	Abstract
12:15-14:15	Lunch
14:15-14:45	Matthias Ehrhardt, University of Bath	Stochastic Optimisation for Large-Scale Inverse Problems
14:55-15:15	Kostas Papafitsoros, Queen Mary University of London	Learned unrolled schemes for estimating spatio-temporally varying regularisation parameters in image reconstruction	Abstract
15:15-15:45	Refreshments
15:45-16:15	Audrey Repetti, Heriot-Watt University	TBA primal-dual plug-and-play algorithm for computational optical imaging (Joint work with C. Garcia, R. R. Thomson, and J.-C. Pesquet)C	Abstract
16:15-16:45	Martin Benning, Queen Mary University of London	Efficient estimation of optimal sampling patterns for MRI	Abstract
16:45-18:00	Discussions and collaboration
18:00-19:30	Workshop Dinner
	THURSDAY 23 MAY 2024
9:15-9:45	Samuli Siltanen, University of Helsinki
9:45-10:15	Tiangang Cui, University of Sydney	Scalable conditional transport maps using tensor trains	Abstract
10:15-10:45	Refreshments
10:45-11:15	Tatiana Bubba, University of Bath	Sparse dynamic tomography regularization using optimal space-time priors	Abstract
11:15-11:45	Matthew Li, MIT	Dimension Reduction for Statistical Inference via (Dimensional)	Abstract
11:45-12:15	Zeljko Kereta, UCL
12:15-14:15	Lunch
14:15-14:45	Tan Bui-Thanh, University of Texas Austin	An autoencoder compression approach for accelerating large-scale inverse problems	Abstract
14:45-15:15	Yiqui Dong, DTU	Sampling strategies in sparse bayesian inference	Abstract
15:15-15:45	Refreshments
15:45-16:15	Markus Holzleitner, University of Genoa	Nonlinear functional regression	Abstract
16:15-16:45	Felix Lucka, CWI Amsterdam	Learning for X-ray Computed Tomography	Abstract
16:45-17:15	Rebecca Willett
	FRIDAY 24 MAY 2024
9:15-9:45	James Nagy, Emory University	Mixed Precision Arithmetic for Large Scale Inverse Problems
9:45-10:15	Mirjeta Pasha, Virginia Tech	Modern Challenges in Large-Scale and High Dimensional Data Analysis	Abstract
10:15-10:45	Refreshments
10:45-11:15	Jakob Sauer Jorgensen, DTU	CUQIpy: Computational Uncertainty Quantification for Inverse Problems in Python	Abstract
11:15-11:45	Marta Betcke, UCL	Learned Stochastic Primal Dual method with applications in subsampled and low dose CT	Abstract
11:45-12:15	Eric de Sturler, Virginia Tech	Streaming Methods for Inverse Problems	Abstract
12:15	End of workshop