Call for Papers (Non-archival)
We welcome up to 5-page workshop papers on topics connecting Information Theory and Generative AI.
Submission & Important Dates
- Submission deadline: Apr. 7, 2026, AoE
- Submission portal: EDAS
Topics of Interest
We invite submissions on (but not limited to) the following topics, organized under two complementary themes.
I. Information Theory for Generative AI
Using information-theoretic principles to understand, analyze, and improve generative AI systems.
-
Generalization, Representation, and Reasoning in GenAI
Information-theoretic analyses of generalization, representation learning, in-context learning, and reasoning in large language and multimodal models. -
Compression, Rate–Distortion, and Efficiency in Generative Models
Rate–distortion–perception tradeoffs, model compression, quantization, and efficient training or inference of generative models. -
Trustworthiness, Robustness, and Security of GenAI
Information-theoretic foundations of watermarking, privacy, robustness to distribution shift or attacks, reliability, and controllability.
II. Generative AI for Information Theory
Leveraging GenAI as a tool or catalyst for advancing information-theoretic research.
-
GenAI-Assisted Theoretical Discovery and Analysis
Conjecture generation, proof assistance, symbolic reasoning, exploration of open problems in information theory, and AI agents for research that support iterative hypothesis testing, literature synthesis, and theory-driven experimentation. -
Learning-Based Methods for Classical IT Problems
Applications of GenAI to coding, compression, inference, estimation, and communication system design. -
Simulation, Modeling, and Algorithmic Design via GenAI
Using generative models to simulate channels, sources, or learning dynamics and to inspire new algorithmic frameworks.
Submission Guidelines
📄 Paper Format
- Authors are encouraged to use LaTeX.
- Please use the official ISIT 2026 paper template.
- Manuscripts must be no longer than 5 pages, with an optional 6th page for references only.
- Fonts must be 10pt or larger.
- Do not adjust template margins.
- Do not include page numbers, headers, or footers.
- Fonts should not be embedded in the PDF.
📎 Supplementary Material
- Additional pages of supplementary material are not allowed in the EDAS submission.
- Authors may include a link to an online preprint (e.g., arXiv or TechRxiv).
- Please list the URL in plain text (do not hyperlink it).
👤 Author Information
- Author affiliations should include an email address for correspondence.
Presentation Format
- Accepted papers will be presented as contributed oral and/or poster presentations.
- The workshop is non-archival.
(When you’re ready, send me the exact wording + deadlines you want on the site, and I’ll lock it in.)