Optimal Randomized Proper Online Learning
En palabras de los autores
We prove that the optimal expected mistake bound of online learning a function class by a randomized proper learning algorithm is , where is the Littlestone dimension of and is the time horizon. Our result improves upon the previously best known bound of given by Daskalakis and Golowich (STOC 2022), and is optimal up to a universal constant for worst-case classes.
Resultado principalEl resumen no menciona limitaciones.
Apareció: lunes, 21 de septiembre. arXiv. Preprint, todavía sin revisión por pares.