The KL-UCB Algorithm for Bounded Stochastic Bandits and Beyond.
Aurélien GarivierOlivier CappéPublished in: COLT (2011)
Keyphrases
- learning algorithm
- computational complexity
- detection algorithm
- improved algorithm
- dynamic programming
- worst case
- times faster
- high accuracy
- experimental evaluation
- np hard
- input data
- expectation maximization
- search space
- preprocessing
- convergence rate
- multi armed bandit
- computational cost
- least squares
- genetic algorithm
- k means
- computationally efficient
- theoretical analysis
- optimization algorithm
- tree structure
- objective function