claim
The researchers derived an instance-specific lower bound on the sample complexity of learning the best action with fixed confidence in online learning with feedback graphs, even when the graph is unknown and stochastic.

Authors

Sources

Referenced by nodes (1)