I'm trying to find an early paper discussing how models, loss functions, and regularizers can all be seen as the same thing. For example, instead of changing the loss function, I can use the standard log loss and instead change the marginal data distribution under the model. Similarly, instead of adding a regularizer to the objective, I can change the model.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.