Bounds on the price of feedback for mistake-bounded online learning
This work provides incremental theoretical improvements to bounds in online learning, relevant for researchers in machine learning theory.
The paper improves worst-case bounds for online learning scenarios, sharpening upper bounds by factors of 2 and 2.41, improving a lower bound by a factor of Θ(ln k), and solving a problem on the price of bandit feedback for multiclass learning.
We improve several worst-case bounds for various online learning scenarios from (Auer and Long, Machine Learning, 1999). In particular, we sharpen an upper bound for delayed ambiguous reinforcement learning by a factor of 2 and an upper bound for learning compositions of families of functions by a factor of 2.41. We also improve a lower bound from the same paper for learning compositions of $k$ families of functions by a factor of $Θ(\ln{k})$, matching the upper bound up to a constant factor. In addition, we solve a problem from (Long, Theoretical Computer Science, 2020) on the price of bandit feedback with respect to standard feedback for multiclass learning, and we improve an upper bound from (Feng et al., Theoretical Computer Science, 2023) on the price of $r$-input delayed ambiguous reinforcement learning by a factor of $r$, matching a lower bound from the same paper up to the leading term.