1,797 followers
Conformal prediction気になるけど成書はやはり高い…Algorithmic Learning in a Random World | SpringerLink https://t.co/ToBgVYa80M
Conformal prediction気になるけど成書はやはり高い…Algorithmic Learning in a Random World | SpringerLink https://t.co/ToBgVYa80M
@MarkTan72526562 @sirbayes Plenty of theory in conformal prediction -> https://t.co/X0DZVmnxsh
An important case is E[Y|X], which is identifiable if (X,Y) are jointly observable. But it's impossible to get "narrow" distribution-free confidence band of E[Y|X] w/ marginal coverage if X is continuous, even if Y is binary; see Vovk's book (Chap. 5) http