Navigating the Unknown: A Workshop on Uncertainties in Machine Learning

Are you a researcher grappling with quantifying uncertainty in your machine learning models? Do you want to connect with leading experts and fellow scientists pushing the boundaries of Uncertainty Quantification (UQ)?

This workshop is designed for researchers and practitioners interested in the vital, and often overlooked, topic of uncertainty in machine learning.  We'll foster intense scientific exchange and discussion on the challenges and promising avenues for applying UQ methods to real-world ML problems.

Workshop Overview

This workshop isn't just about theory. We aim to bridge the gap between cutting-edge research and practical application.  Expect engaging discussions with practitioners, a focused exploration of current UQ methods, and opportunities to collaborate on future research.

Dates: January 13-15, 2026
Location: Helmholtz-Zentrum Dresden-Rossendorf, Bautzner Landstrasse 400, 01328 Dresden, Germany

What to Expect:

• Lunch-to-Lunch Workshop (Jan 13-14, 2026): Two full days of presentations, discussions, and networking focused on the latest advances in UQ for ML.  We will cover a broad range of topics including (but not limited to)
  • Bayesian Neural Networks   
  • Conformal Prediction
  • Gaussian Processes
  • Laplace Approximation
  • Ensemble Methods for UQ
  • Robustness and Adversarial Training
  • Uncertainty-Aware Deep Learning
  • Applications of UQ in science domains (e.g., energy, climate, physics, health, ...)
• Optional Hackathon (Jan 14-15, 2026):  Put theory into practice! We'll be hosting an optional hackathon focused on building a resource for the UQ community.  The goal is to collaboratively develop a decision tree to help researchers navigate the landscape of available UQ methods based on their specific problem characteristics.  We aim to produce a manuscript based on the hackathon results.
• Intense Scientific Exchange:  The workshop will prioritize lively discussion and collaboration, creating an environment for genuinely impactful knowledge sharing.
• Practitioner Insights: Hear directly from individuals applying UQ methods in industry and research, learning about their successes and challenges.
• Keynote Speaker:  Stay tuned! We are in the process of confirming an exciting keynote speaker who is a leader in the field.  News will be announced shortly.

Call for Abstracts!

Share your work with the community!

We are soliciting abstracts for presentations on all aspects of uncertainty quantification in machine learning.  We encourage submissions describing novel methods, applications, theoretical analyses, and practical challenges.

Important Dates:

• Abstract Submission Deadline: October 24, 2025
• Notification of Acceptance: November 2, 2025

Abstract Guidelines

• Abstracts should be no more than 300 words.
• Please clearly state the problem, approach, and key results.
• Submit abstracts via our submission form

Who Should Attend?

This workshop is ideal for:

*   Researchers in machine learning, statistics, and related fields.
*   Practitioners and Engineers applying ML to real-world problems where uncertainty is critical.
*   Graduate students interested in learning about UQ in ML.
*   Anyone seeking to understand and address the challenges of reliable machine learning and trustworthiness with UQ in ML.

Registration

Registration details, including possible fees and accommodation information, will be available starting November 2, 2025.

Depending on the received abstracts, we reserve the right to adopt the workshop programme accordingly.

Organizers

• Peter Steinbach, Helmholtz AI / HZDR

Contact Us

For any questions or inquiries, please contact us at [Workshop Email Address].

We look forward to seeing you in January 2026!