Researchers have introduced a new bandit problem framework designed for smooth functions on graphs, applicable to online learning tasks like content-based recommendation. The proposed algorithms aim to minimize cumulative regret by considering an 'effective dimension' of the graph, which is shown to be small in real-world scenarios. Experiments demonstrate that user preferences for thousands of items can be effectively learned from a limited number of node evaluations. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
RANK_REASON The submission is an academic paper on arXiv detailing a new machine learning framework.
