TAG in AI Robustness, Explainability and Safety 2026

A Workshop at the Joint Math Meetings (JMM 2026), Seattle, WA, January 4, 2026

Answering questions around the safety, robustness, and explainability of AI models is becoming increasingly critical. Mathematics helps us understand AI failure modes and make AI more transparent and reliable. This special session features mathematics research that analyzes and addresses AI assurance concerns, showcasing areas such as algebraic geometry, probability theory, and computational topology which provide the insights required for AI systems to meet the needs of real-world applications.

Organizers:

Scott Mahan, Pacific Northwest National Lab

Eric Yeats, Pacific Northwest National Lab;

Henry Kvinge, Pacific Northwest National Laboratory;

Tim Doster, Pacific Northwest National Lab;

Alexander Cloninger, UCSD

Mathematics for AI Robustness, Explainability, and Safety

Location: Room 102A, Walter E. Washington Convention Center

Motivating coherence-driven inference via sheaves

8:00 a.m.

Steve Huntsman

Explicit loss minimizers and geometric generalization bounds in deep networks

8:30 a.m.

Thomas Chen

Curvature Tuning: Provable Training-free Model Steering From a Single Parameter

9:00 a.m.

Randall Balestriero

Poisoning Large Language Models with Model Editing

9:30 a.m.

David Shriver, Keltin Grimes, Marco Christiani, Marissa Connor

Retraining Emulation: A General Framework for Machine Unlearning

10:00 a.m.

Yiran Jia, Eric Yeats, Scott Mahan

The Measure of Deception: An Analysis of Data Forging in Machine Unlearning

10:30 a.m.

Rishabh Dixit

Detecting Collateral Damage in Unlearning for Diffusion-Based Image Generation Models

11:00 a.m.

Aaron Jacobson, Scott Mahan

Machine Unlearning via Information Theoretic Regularization

11:30 a.m.

Shizhou Xu, Thomas Strohmer

Page updated

Google Sites

Report abuse