New FTPL Policy Achieves Near-Linear-Time Learning for Weakly Observable Graph Bandits
Researchers propose a Follow-the-Perturbed-Leader policy for online learning with directed weakly observable feedback graphs, achieving best-of-b... (truncated in report)
Read the full story at arXiv stat.ML ↗
Timeline · 1 report
- 2026-11-04 05:00 · arXiv stat.ML
Nearly Linear-Time Follow-the-Perturbed-Leader for Weakly Observable Graph Bandits