Fast UCB-type algorithms for stochastic bandits with heavy and super heavy symmetric noise.
Yuriy DornAleksandr KatrutsaIlgam LatypovAndrey PudovikovPublished in: CoRR (2024)
Keyphrases
- multi armed bandit
- data structure
- computational complexity
- orders of magnitude
- computationally efficient
- noisy data
- worst case
- times faster
- learning algorithm
- data mining techniques
- data sets
- computational efficiency
- evolutionary algorithm
- dynamic programming
- lower bound
- bayesian networks
- convergence rate
- noise level
- restrictive assumptions
- bandit problems