This workshop focuses on Machine Learning for Algorithms, interpreted broadly as using learned models and data to design more efficient algorithms with improved runtime, memory, or solution quality, while aiming for provable guarantees. The workshop will cover two active and complementary threads in this area, as well as the connections between them:
The area has grown from early empirical successes into a well-established research direction across multiple communities, supported by a steady stream of publications and recent activity across online and streaming algorithms, graph problems, clustering, and data structures (check this repository). This workshop aims to bring the theory community up to speed on recent major developments and to lay out major open questions in the area.
Tutorials will be given by the organizers. The following invited speakers have confirmed their participation:
A few additional speakers will be announced soon.
Each day consists of a 2-hour session. Detailed timings, titles, and abstracts will be posted as they are finalized.
We invite poster submissions on recent or ongoing work in machine learning for algorithms, spanning both data-driven algorithm design and learning-augmented algorithms, as well as adjacent topics. The poster session on Day 2 is intended to encourage informal discussion and new collaborations across the community.
To submit a poster, please fill out the submission form:
Both published and in-progress work are welcome. The submission deadline and notification schedule will be announced shortly. Questions can be directed to vakilian [at] vt [dot] edu.
The workshop is part of STOC 2026 / TheoryFest 2026, held June 22–26 in Salt Lake City, Utah. Attendance follows STOC registration; please refer to the STOC 2026 website for registration, venue, and travel information.