Algorithm Selection as a Bandit Problem with Unbounded Losses
Matteo GaglioloJürgen SchmidhuberPublished in: CoRR (2008)
Keyphrases
- cost function
- objective function
- learning algorithm
- experimental evaluation
- improved algorithm
- k means
- computational complexity
- preprocessing
- significant improvement
- np hard
- selection algorithm
- detection algorithm
- high accuracy
- linear programming
- estimation algorithm
- similarity measure
- classification algorithm
- computationally efficient
- particle swarm optimization
- simulated annealing
- worst case
- artificial neural networks
- theoretical analysis
- computational cost
- times faster
- convergence rate
- dynamic programming
- feature space
- selection strategy
- single pass