An optimal algorithm for the Thresholding Bandit Problem.
Andrea LocatelliMaurilio GutzeitAlexandra CarpentierPublished in: CoRR (2016)
Keyphrases
- dynamic programming
- optimal solution
- experimental evaluation
- preprocessing
- k means
- worst case
- theoretical analysis
- learning algorithm
- times faster
- detection algorithm
- globally optimal
- linear programming
- closed form
- high accuracy
- matching algorithm
- convergence rate
- recognition algorithm
- particle swarm optimization
- cost function
- significant improvement
- search space
- thresholding algorithm
- locally optimal
- multiscale
- space complexity
- exhaustive search
- reinforcement learning
- data structure
- computational complexity
- genetic algorithm
- path planning
- optimization algorithm
- np hard
- expectation maximization
- computational cost
- denoising