Dear reviewers$
We are grateful to hear that all the reviewers recognized that our submission contains significant contribution to the safe screening literature. We do not have strong rebuttals to any of the reviewers comments. Here are some clarifications to the comments and answers to the questions.

 < To Reviewer 1 > 

Comment: All the assumptions should be explicitly mentioned in theorems.

Our answer: We totally agree with the reviewer. In the final version, we will explicitly mention all the assumptions (which have been separately presented in problem formulation part in section 2 in the current manuscript) within theorem parts.

Comment: Quantitative results on time-savings by simultaneous screening and safe keeping should be presented.

Answer: In the paper, we mainly presented results on screening rates rather than time-saving because the latter highly depends on the choice and implementation of the optimization solvers. In final version, we will put some quantitative results on time saving in the appendix.

 < To Reviewer 2 > 

Comment: How was the performance of simultaneous screening for LP-based SVM?

Our answer: We observe relative advantages of simultaneous screening also for LP-based SVM. We will put some of those results in the appendix of final version.

Comment: Applicability to one-class SVM

Our answer: Although we have not particularly considered one-class SVM, we will discuss possibilities of extensions to other problem setups such as one-class SVM.

 < To Reviewer 3 > 

Comment: Clear explanation on safe keeping is desired.

Our answer: In the appendix of the final version, we will present algorithmic statement on how safe keeping can be implemented and used.

 < To all reviewers > 

Finally, we would like to thank all the reviewers for pointing out typos and additional references. We will fix all the typos, and add the suggested references in the final version. We would also thank for several suggestions for improving the presentation. We will reflect these suggestions in the final version.