Contextual bandits with surrogate losses: Margin bounds and efficient algorithms.
Dylan J. FosterAkshay KrishnamurthyPublished in: CoRR (2018)
Keyphrases
- regret bounds
- lower bound
- online learning
- upper bound
- linear regression
- contextual information
- worst case
- maximum margin
- multi armed bandit
- objective function
- rademacher complexity
- data sets
- context sensitive
- stochastic systems
- training error
- upper and lower bounds
- high level
- data dependent
- context dependent
- generalization error
- support vector
- decision boundary
- tight bounds
- least squares
- np hard
- pairwise
- risk bounds
- multi armed bandits