b59307fdacf7b2db12ec4bd5ca1caba8-AuthorFeedback.pdf

We thank the reviewers for the thoughtful comments. We will address ALL comments in our final version. Reviewer 1: We do not claim that we improve a standard model for image classification. However, our sample complexity speedup performance is most likely to be superior. Weights sharing across active rounds - We train the network from scratch on each active learning round. It might not always be the case that more data induce deeper (more complicated) model... - We think that this situation Random querying baseline - we agree that random querying (i.e.