2,190 followers
@roydanroy @_vaishnavh @zicokolter my statement about SRM-like bounds depends on the hypothesis class / specific bound. I was thinking about the class of all Boolean functions on a finite domain, assuming realizability, and using a bound of the form Prelim