Embedded Files
Program at a Glance
Program at a Glance
Monday, February 2
Monday, February 2
09.00 Registration opens
09.30 - 11.15 Pre-Workshop Tutorial on Uncertainty
11:15 Coffee Break
11.45 - 12.00 Welcome & Opening
12.00 - 13.00 Keynote
13:00-14:00 Lunch Break
14.00 - 15.30 Presentations - Session 1
15.30 - 16.00 Coffee Break
16.00 - 18:00 Presentations - Session 2
19:00 Conference dinner - restaurant Väike Till (Kastani 42).
Tuesday, February 3
Tuesday, February 3
09.00 - 10.00 Invited talk
10.00 - 11.30 Flash talks,
Poster session and coffee
11.30 - 13.00 Presentations - Session 3
13.00 - 14.00 Lunch Break
14.00 - 15.00 Presentations - Session 4
15.00 - 16.00 Keynote
16.00 - 17.30 Flash talks,
Poster session and coffee
Social mixer
Wednesday, February 4
Wednesday, February 4
09.00 - 10.30 Presentations - Session 5
10.30 - 11.00 Coffee Break
11.00 - 13.00 Presentations - Session 6
Closing
Optional program
Keynote speakers
Keynote speakers
Prof. Willem Waegeman is a Professor at Ghent University in the Department of Data Analysis and Mathematical Modelling, within the Faculty of Bioscience Engineering. He is leading the BioML research unit (www.bioml.ugent.be). His research focuses on machine learning and data science, particularly uncertainty quantification, multi-target and other complex prediction problems, deep learning and applications in “omics” data analysis.
“Disentangling aleatoric and epistemic uncertainty in ML: challenges and opportunities”
Abstract: Given the increasing use of machine learning (ML) models for decisions that directly affect humans, it is essential that these models not only provide accurate predictions but also offer a credible representation of their uncertainty. Recent advances have led to probabilistic models capable of disentangling two types of uncertainty: aleatoric and epistemic. Aleatoric uncertainty is inherent to the data and cannot be eliminated, while epistemic uncertainty is related to the ML model and can be reduced with better modeling approaches or more data. In this talk I will elaborate on the limitations and opportunities of uncertainty disentanglement in explaining why an ML model fails to deliver accurate predictions.
Ilja Kuzborskij is a research scientist at the Foundations team at Google DeepMind in London. He obtained his Ph.D. from the Swiss Federal Institute of Technology Lausanne (EPFL) in 2018 and did postdoctoral research at the University of Milan with Nicolò Cesa-Bianchi. His work mainly focuses on problem-dependent theories of
generalization, uncertainty estimation, and related topics such as concentration inequalities.
"Distinguishing Between Aleatoric and Epistemic Uncertainty in Large Language Models"
Invited speaker
Invited speaker
Arun Kumar Singh is a Professor at Institute of Technology, University of Tartu. Earlier he has been a postdoc at Ben Gurion University, Israel, Nanyang Technological University, Singapore and Tampere University of Technology, Finland. His research interests lie in the algorithmic foundation of robotics in the context of motion planning and control and machine learning. The current research threads include optimization for robotics and learning, multi-agent navigation, autonomous driving, shared control, motion planning under uncertainty.
"Leveraging Predictive Uncertainty for Reliable Model-Based Planning and Control"
Model-based planning and control stands as a dominant paradigm for autonomous decision-making in robotics. In this context, the "model"—whether describing robot dynamics or environmental interactions—is increasingly learned from data. However, data-driven models are prone to errors. In this talk, I will argue for the necessity of capturing the uncertainty associated with these learned models and demonstrate how these estimates can be utilized to achieve robust planning and control. I will highlight recent advancements, including work from my own group, on learning uncertainty in a task-aware fashion. To ground these theoretical discussions, I will present applications in monocular image-based navigation and autonomous driving.
Talks
Talks
Session 1 (Monday, February 2) - Uncertainty methods
Mira Jürgens – Selective prediction with application in molecular structure retrieval
Daniel Romero Alvarado – What should an AI assessor optimise for?
Sebastian Jimenez – Uncertainty Quantification in Geospatial Modeling
Session 2 (Monday, February 2) - Data uncertainty and conformal prediction
Cornelia Gruber – Quantifying Ambiguity in Data Sets
Fanyi Wu – Bayesian Conformal Prediction as a Decision Risk Problem Using Bayesian Quadrature
Soundouss Messoudi – SPACR: Single-Pass Adaptive Training of Uncertainty-Aware Conformal Regressors
Session 3 (Tuesday, February 3) - Autonomous driving and Decision making
Ziliang Xiong – Collision Risk Estimation via Loss Prediction in End-to-End Autonomous Driving
Basant Sharma – MMD-OPT: Maximum Mean Discrepancy-Based Sample-Efficient Collision Risk Minimization for Autonomous Driving
Javier Fumanal Idocin – Dynamic Feature Selection with Uncertainty Quantification
Session 4 (Tuesday, February 3) - Robustness Quantification
Jasper De Bock – Robustness Quantification: using imprecise probabilities to assess the reliability of probabilistic classifiers
Adrian Detavernier – Robustness Quantification: using imprecise probabilities to assess the reliability of probabilistic classifiers (PART II)
Rodrigo Ferrari Lucas Lassance – Robustness Quantification: using imprecise probabilities to assess the reliability of probabilistic classifiers (PART III)
Session 5 (Wednesday, February 4) - Labels and quantification
Vitor Martin Bordini – Cautious Self-Learning via Labelwise Uncertainty Quantification
Jan Mielniczuk – Prior shift estimation for positive unlabeled data through the lens of kernel embedding
Blagovest Gospodinov – Are soft labels worth the trouble
Session 6 (Wednesday, February 4) - Uncertainty and Optimization
Jelle Hüntelmann – Towards Uncertainty Calibration: ARC Revisited
Jan Timko – Optimizing stock price forecasting: a hybrid approach using fuzziness and automated machine learning
Kārlis Freivalds – Solving Boolean Satisfiability by Repeated Knowledge Accumulation
Poster presentations
Poster presentations
Poster Session A
Daniel Romero Alvarado – What should an AI assessor optimise for?
Sebastian Jimenez – Uncertainty Quantification in Geospatial Modeling
Sam Goring – On the QUEST for uncertainty quantification methods for regression
Vladyslav Fediukov – Uncertainty-aware multi-fidelity Gaussian processes
Jelle Hüntelmann – Towards Uncertainty Calibration: ARC Revisited
Jaime de Miguel – A Methodology for Exploring the Interplay Between Battery Design, Data Uncertainty, and Planning Horizon in Multi-Stage Optimization.
Lucas Ferrando Plo – Uncertainty quantification with conformal predictive systems in regression of a molecular property.
Fabian Denoodt – How many samples does my Bayesian Neural Network need? A Confidence Sequence Perspective
Matias Valdenegro-Toro – Measuring Uncertainty Disentanglement Error in Classification
Timo Löhr – Efficient Credal Prediction through Decalibration
Fanyi Wu – Bayesian Conformal Prediction as a Decision Risk Problem Using Bayesian Quadrature
Soundouss Messoudi – ConfSort: Conformal Models for Plastic Waste Sorting
Joonas Järve – Probability Density from Latent Diffusion Models for Out-of-Distribution Detection
Poster Session B
Mark Fišel – Automatic Generation of Under-resourced Languages and the Impact of Uncertainty
Larkin Liu – Optimal Incentivization for the Multi-Follower Principal Agent Stackelberg Game
Basant Sharma – Sample Efficient Risk Estimation Through Maximum Mean Discrepancy
Aurel Davy TCHOKPONHOUE – On the value of uncertainty quantification in deep learning based breast cancer molecular subtype classification
Blagovest Gospodinov – Are soft labels worth the trouble
Jan Timko – Optimizing stock price forecasting: a hybrid approach using fuzziness and automated machine learning
Chandan Kumar Sah – Uncertainty Quantification in Machine Learning for Responsible AI
Laura Leja – Shaping Flames with Differentiable Physics Simulations
Guntis Vilnis Strazds – Harnessing Foundation Models to Generate and Verify Robot Control Policies
Rebeka Birzina – Test Time Scaling Methods For Neural SAT Solve
Linda Barbare – Analysis of Latvian Spoken Songs Using Artificial Intelligence
Javier Fumanal Idocin – Dyanmic Feature Selection with ncertainty Quantification
Xabier Gonzalez Garcia – Uncertainty Quantification for Capacities based on Integral Imprecise Probability Metrics
Jasper De Bock, Adrián Detavernier, Rodrigo Lassance – Robustness Quantification: a new paradigm for assessing the reliability of probabilistic classifiers
Ziliang Xiong – Towards Safer Autonomous Systems: Uncertainty Quantification for Regression
Novin Shahroudi – Aligning the Evaluation of Probabilistic Predictions with Downstream Value
Google Sites
Report abuse