Online Algorithms with Unreliable Guidance

In three linesNew arXiv paper introducing OAG (Online Algorithms with Unreliable Guidance), a model for ML-augmented online decision-making separating predictive and algorithmic components. Presents DTB (drop-or-trust-blindly) compiler converting standard online algorithms into learning-augmented versions. Demonstrates optimal guarantees on bipartite matching, caching, and uniform metrical task systems.

Summary generated by Claude — human-verified