Non-Uniform Multiclass Learning with Bandit Feedback

–Neural Information Processing Systems 

We study the problem of multiclass learning with bandit feedback in both the i.i.d.