Keynote Speaker
Arthur Gretton
University College London and Google DeepMind
Foundations
ICML 2026 Workshop
ICML 2026 Workshop on Hypothesis Testing
Advancing the foundations and applications of hypothesis testing in machine learning
A workshop bringing together researchers developing modern testing methodology and applying it across machine learning, including robustness, distribution shift, security, medicine, and LLM evaluation.
- 6 invited speakers
- Contributed talks and posters
- Panel discussion
Why this workshop
Hypothesis testing remains a core tool for reliable machine learning.
The workshop highlights both the enduring foundations of statistical testing and the new demands created by modern machine learning.
Hypothesis testing remains central to scientific inquiry and continues to play a foundational role in machine learning, where empirical claims often depend on reliable comparisons, validation procedures, and uncertainty-aware decision making.
Modern machine learning introduces new challenges for testing methodology, including adaptive analysis, distribution shift, model complexity, safety-critical deployment, and the evaluation of increasingly powerful systems such as large language models.
This workshop aims to bridge foundational advances in testing with practical applications across machine learning, creating a common forum for researchers working on both rigorous methodology and real-world impact.
Topics
Foundations and applications
The program spans methodological advances in testing as well as high-impact machine learning domains where rigorous testing is essential.
Foundations of Testing Methods
- 01Anytime-valid inference and e-values
- 02Adaptive hypothesis testing and differential privacy
- 03Learned-representation tests
- 04Conditional independence and homogeneity testing
- 05Foundational limits and impossibility results
Impacts and Applications
- 01A/B testing
- 02Domain adaptation and model selection
- 03Out-of-distribution detection and anomaly detection
- 04Validation of ML assumptions and explanations
- 05Membership inference and adversarial testing
- 06Medical subgroup shift detection
- 07LLM-related testing problems
Keynote speakers
Leading perspectives across foundations and applications
The invited program brings together researchers spanning testing theory, security, medicine, and learning-based decision systems.
Keynote Speaker
Yao Xie
Georgia Institute of Technology
Foundations
Keynote Speaker
T
TBA
TBA
Foundations
Keynote Speaker
Bo Li
University of Illinois Urbana-Champaign and Virtue AI
Security
Keynote Speaker
T
TBA
TBA
Medicine
Keynote Speaker
Yisong Yue
Caltech and Latitude AI
Robotics / Control
Call for Papers
We invite submissions on modern hypothesis testing in machine learning.
We welcome submissions on the theory, methodology, and practice of hypothesis testing in machine learning. The workshop is designed to bring together researchers developing new testing methods and researchers applying them in important ML domains. Submissions should follow the ICML 2026 format.
Topics of interest
- Theoretical advances in modern hypothesis testing
- Testing under adaptivity, dependence, and privacy constraints
- Evaluation under robustness challenges and distribution shift
- Testing problems arising in security, medicine, and LLM systems
Submission types
- Contributed short papers (4 pages)
- Contributed long papers (8 pages)
Submission details
- Submission deadline: 10 May 2026
- Notification date: 26 May 2026
- Camera-ready deadline: 17 June 2026
- Workshop date: July 10 or 11 (TBA)
Presentation format
Accepted submissions may be presented as posters and some of them will be selected for contributed talks (i.e., orals).
Schedule
Single-day workshop program
The program is organized around foundational advances, domain-facing applications, and a closing panel discussion.
Session 1: Foundation of Testing
7 items09:00-09:50
Opening and Keynote 1
09:50-10:00
Contributed Talk 1
10:00-10:45
Keynote 2
10:45-10:55
Contributed Talk 2
10:55-11:10
Morning Tea Break
11:10-11:55
Keynote 3
11:55-12:05
Contributed Talk 3
Session 2: Important Sectors Related to Testing
5 items13:30-14:15
Keynote 4
14:15-14:25
Contributed Talk 4
14:25-15:10
Keynote 5
15:10-15:50
Afternoon Tea Break / Poster Session
15:50-16:35
Keynote 6
Session 3: Panel Discussion
2 items16:35-16:55
Panel Discussion
16:55-17:00
Closing Remarks
Organizers
Workshop organizers
The organizing team (alphabetical order by surname) brings together expertise from academia and industry across statistics and machine learning.
Organizer
Xiuyuan Cheng
Organizer
Duke University
Researcher working at the intersection of machine learning, statistics, and modern testing methodology.
Organizer
Feng Liu
Organizer
University of Melbourne
Researcher in trustworthy machine learning, hypothesis testing, distribution shift, and statistical foundations for robust evaluation.
Organizer
Lester Mackey
Organizer
Microsoft Research / Stanford affiliation
Researcher in scalable inference, adaptive data analysis, and reliable machine learning systems.
Organizer
Shayak Sen
Organizer
Snowflake
Researcher focused on statistical methodology, experimentation, and practical decision-making systems.
Organizer
Danica J. Sutherland
Organizer
University of British Columbia
Researcher in machine learning and statistics with interests in testing, kernels, and reliable model evaluation.
Organizer
Nathaniel Xu
Organizer
University of British Columbia
Researcher working on statistical machine learning and practical testing problems in modern data settings.
Attend / Logistics
Attendance and practical information
Venue, registration, and contact details will be updated here as ICML 2026 planning progresses.
The workshop is intended for researchers and practitioners interested in the foundations and applications of hypothesis testing across modern machine learning.
Additional details about attendance, access, and conference-specific policies will be added here once ICML 2026 logistics are finalized.
Please check back for updates on venue assignment, registration procedures, and any workshop-specific instructions.
Workshop logistics
- Venue: The COEX Convention & Exhibition Center, Seoul, South Korea
- Workshop date: July 10 or 11 (TBA)
- Registration information: Follow the ICML 2026 registration page.
- Contact: fengliu.ml@gmail.com
FAQ
Common questions
A few common questions about attendance, submissions, and presentation format.