Spectral bandit algorithms for graph-smooth payoffs achieve regret linear or sublinear in an effective dimension that is small on real graphs, allowing preference learning from tens of evaluations on thousands of items.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Spectral bandits for smooth graph functions
Spectral bandit algorithms for graph-smooth payoffs achieve regret linear or sublinear in an effective dimension that is small on real graphs, allowing preference learning from tens of evaluations on thousands of items.