Pierre Ménard
Published 2019-05-20Version 1
We present a new algorithm based on an gradient ascent for a general Active Exploration bandit problem in the fixed confidence setting. This problem encompasses several well studied problems such that the Best Arm Identification or Thresholding Bandits. It consists of a new sampling rule based on an online lazy mirror ascent. We prove that this algorithm is asymptotically optimal and, most importantly, computationally efficient.
Comments: 21 pages, 1 figure
On the Complexity of Best Arm Identification in Multi-Armed Bandit Models
Best Arm Identification in Stochastic Bandits: Beyond $β-$optimality
On Uniformly Optimal Algorithms for Best Arm Identification in Two-Armed Bandits with Fixed Budget
![QR code for arXiv:1905.08165 [stat.ML]](http://oss.arxitics.com/images/favicon.svg)