Skip to yearly menu bar Skip to main content


ICML 2026 Call for Tutorials

 

ICML 2026 will be held in Seoul, South Korea, with tutorials taking place on July 6, 2026. The ICML tutorial program aims to provide high-quality, in-depth presentations on topics of broad interest to the machine learning community, covering both foundational advances and emerging research directions.

 

Mandatory Tutorial Topics

Following an open call for community nominations, ICML 2026 will only consider tutorial proposals that directly address one of the following topics:

  1. Beyond-Transformer Sequence Models (e.g., state-space models, S4, Mamba, and related architectures)
  2. Diffusion Models: Quantitative and Theoretical Understanding
  3. Safety, Machine Unlearning, Watermarking, and Fingerprinting
  4. Deep Learning and Deep Reinforcement Learning: Theory and Applications
  5. LLM Post-Training and Test-Time Training
  6. Theorem Proving and Formal Methods with Lean

Proposals that do not clearly fall within one of these topics will not be considered.

 

Tutorial Format

  • Tutorials will be in-person and 2.5 hours long, including time for questions.
  • Each tutorial may have up to two presenters. Each presenter will be offered a complementary full conference registration. 
  • Tutorials should be accessible to a broad ICML audience while providing sufficient technical depth.
     

Submission Guidelines

Tutorial proposals must be two pages and should clearly describe:

  • The tutorial’s scope and learning objectives
  • The specific content to be covered
  • The intended audience and prerequisites
  • The presenter(s) and their affiliation(s)
  • Any planned supplementary materials (e.g., slides, notes, code)

Submissions will be handled through an online submission form (details to be announced).

 

Important Dates

  • Submission deadline: Friday, February 6, 2026 (AOE)
  • Notification of decisions: Early March 2026
     

Review Process

All proposals will be reviewed by the ICML 2026 Tutorial Chairs. Selection will be based on relevance to the specified topics, clarity of presentation, expected impact on the community, and balance across topics and presenters.