Online Learning in Case of Unbounded Losses Using the Follow Perturbed Leader Algorithm
Vladimir V. V'yuginPublished in: CoRR (2010)
Keyphrases
- online learning
- improved algorithm
- learning algorithm
- times faster
- objective function
- dynamic programming
- search space
- experimental evaluation
- expectation maximization
- maximum likelihood
- computational complexity
- significant improvement
- neural network
- selection algorithm
- detection algorithm
- optimal solution
- high accuracy
- data sets
- cost function
- simulated annealing
- mobile robot
- search algorithm
- particle swarm optimization
- preprocessing
- classification algorithm
- similarity measure
- original data
- k means