Search papers, labs, and topics across Lattice.
University of California
1
0
1
KL-regularization in multi-armed bandits provably achieves near-optimal regret, scaling linearly with the number of arms, a significant improvement over classical results.
