For deep learning, are there any theoretical or experimental criterions for task complexity?

For example, we can say roughly, within a set of tasks, tasks that take more gradient updates to achieve the same loss (assuming the losses are on the same scale) are more complex. But how do we talk about this idea rigorously?


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.