Towards Problem-dependent Optimal Learning Rates
Yunbei Xu Assaf Zeevi
Columbia University Columbia University
New York, NY 10027 New York, NY 10025
yunbei.xu@gsb.columbia.edu assaf@gsb.columbia.edu
Abstract
We study problem-dependent rates, i.e., generalization errors that scale tightly
with the variance or the effective loss at the "best hypothesis." Existing uniform
...
附件列表