Regular Paper Track
Up to 8 pages, excluding references and appendices
Machine Learning/CORE Unranked
XAI4Science 2026
XAI4Science: Knowledge Discovery and Trust through Interpretable Foundation Models
XAI4Science 2026 is a NeurIPS workshop focused on advancing explainable artificial intelligence (XAI) for scientific applications, particularly in weather and climate forecasting using foundation models. The workshop aims to address transparency issues in ML-based models by fostering collaboration between machine learning researchers, climate scientists, and policy experts to develop physically consistent, interpretable, and trustworthy AI systems.
Key deadlines
Verified September 9, 2026Full paper
September 6, 2026
AoE
Conference timeline
Submission and decisions
Full paperKey deadline
September 6, 2026 · AoE
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Tiny Paper Track
Up to 5 pages, excluding references and appendices
Research areas in scope
01
What the Workshop Covers
Ante-hoc interpretability: Self-explainable architectures and inductive biases for Foundation ModelsPost-hoc attribution & evaluation: Attribution, probing, and mechanistic interpretability methods, including rigorous evaluation of their faithfulnessPhysics-consistent explanations: Benchmarks and methods for validating explanations against known physical laws and causal structure
Policies worth checking twice
- Submissions are non-archival and do not publish proceedings
- Submissions can be concurrently or subsequently submitted to other venues
- Submissions must be made via OpenReview
- Each submission must nominate at least one author to serve as a reviewer (reciprocal review rule)
- Accepted works are expected to be presented in person; remote presentations permitted only in exceptional cases
- No Paper Checklist is required for workshop submissions
- Submissions must use the NeurIPS 2026 LaTeX template
- Unlimited pages are allowed for references and appendices in the same PDF as the main paper
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026