Fwd: Guidelines for DL and CNN papers

[Anderson Rocha]

Dear Recodists,

Everyone thinking about submitting a paper to T-IFS using Deep Learning must be AWARE of the new guidelines. This might also be enforced in other IEEE journals soon, so pay careful attention.

Bests,

-- Anderson

(Currently on a Sabbatical Leave in the 
Rapid-Rich Object Search Lab. (ROSE), 
Nanyang Technological University (NTU), Singapore)

************************************************
Prof. Dr. Anderson Rocha
Associate Dean, Institute of Computing
UNIVERSITY OF CAMPINAS, SP - BRAZIL
Digital Forensics and Machine Intelligence
<http://www.ic.unicamp.br/~rocha>
************************************************

---------- Forwarded message ----------
From: Mauro Barni
Date: Mon, Mar 27, 2017 at 6:48 PM
Subject: Guidelines for DL and CNN papers
To: TIFS...@listserv.ieee.org

Dear T-IFS EB members,

thanks to the hard work of the ad-hoc group we formed a while ago, we were able to prepare a document with the guidelines for papers dealing with CNN (and deep learning in general). The guidelines were prepared by having in mind steganalysis and multimedia forensics applications, but the

​y​

are applicable to other fields as well.

I presented the guidelines to the SPS pub board meeting held at the beginning of March. Everybody recognised that the guidelines are written extremely well and we have been given the permission to use them. It took some time, but I finally managed to publish the guidelines on-line. They are accessible through T-IFS web page at the end of the section regarding result reproducibility: http://signalprocessingsociety.org/publications-resources/ieee-transactions-information-forensics-and-security. You can also access the document with the guidelines directly from here: http://signalprocessingsociety.org/sites/default/files/uploads/publications_resources/docs/Guidelines__deep_learning_submissions.pdf.

Now the difficult part comes: we need to apply these guidelines while reviewing CNN and DL papers submitted to T-IFS. To do so, I suggest to proceed as follows:

1. You download the guidelines and get familiar with them

2. Optional but strongly recommended: for each submission I assign to you, you go through the paper yourself and if you find that the paper deviates STRONGLY from the guidelines you inform the authors about the new guidelines asking them to prepare a new version of the paper. If they agree, you ask Allison to rescind the submission waiting for a new version of the paper. This process DOES NOT COUNT AS A REJECTION. Also it is necessary that this check is done in a few days.

3. If the paper passes step 2, you inform the reviewers who agreed to review the paper about our new policy, you pass them the guidelines and ask them to pay particular attention to verify that the paper complies with the guidelines. 

4. In case you finally reject a paper because it is not compliant with the guidelines, you must clearly explain the problem to the authors, and send them the guidelines so that they can follow them in subsequent submissions. Of course, you should do the same with papers for which you suggest a major revision if as part of the revision you ask them to comply with the guidelines.

I know it will take some time before this “message” is widely spread in our community, but I truly believe this is the right way to go. 

Please let me know if you have any doubts about the above procedure.

Best

Mauro