Algorithms with Predictions

About the Workshop

The rapidly-growing area of learning-augmented algorithms, also known as algorithms with predictions, provides a principled framework for algorithm design that integrates imperfect prior information, most notably machine-learned predictions, while maintaining rigorous, provable performance guarantees. Its central goal is to translate advances in modern AI into reliable algorithmic improvements: algorithms should benefit from accurate predictions to achieve near-optimal performance, yet remain robust and reliable when predictions are inaccurate. This perspective raises fundamental questions at the interface of theoretical computer science and AI about how to represent and evaluate predictive information and how to integrate it into algorithmic decision-making in a controlled way. 

This workshop aims to bring together researchers across AI/ML and algorithms to discuss recent progress and identify emerging directions and applications.

List of relevant topic areas

• Algorithm design with predictions

• Mechanism design with predictions

• Prediction models

• Error notions

• Consistency-robustness tradeoffs

• Generative AI/LLMs in algorithm design

• Submission deadline: May 15, 2026

• Notification of acceptance: June 1, 2026

• One-day Workshop: August 15, 2026

The workshop takes place at IJCAI-ECAI 2026 in Bremen, Germany. This will be a one day fully in-person meeting consisting of technical sessions, two keynote speakers, and an open problem session aimed at discussion and fostering collaborations.