Probabilistic Sequential Shrinking: A Best Arm Identification Algorithm for Stochastic Bandits with Corruptions.
Zixin ZhongWang Chi CheungVincent Y. F. TanPublished in: ICML (2021)
Keyphrases
- preprocessing
- learning algorithm
- optimization algorithm
- k means
- segmentation algorithm
- worst case
- monte carlo
- significant improvement
- experimental evaluation
- computational cost
- stochastic approximation
- detection algorithm
- decision trees
- convergence rate
- times faster
- matching algorithm
- improved algorithm
- high accuracy
- computational complexity
- theoretical analysis
- computationally efficient
- em algorithm
- data sets
- probabilistic model
- dynamic programming
- path planning
- np hard
- search space
- reinforcement learning
- parallel version
- context free parsing