<div dir="ltr"><div><br></div><div>Thank you Juyang,</div><div><br></div><div>" Please suggest what we should do, as much resource is being wasted."</div><div><br></div><div>A suggestion: Have a student or a group of students redo the studies (at least those for which the code and data are available) and publish the correct performance results. </div><div><br></div><div>Write down: The original study reported the performance of A. The true performance is B. And B < A.</div><div><br></div><div>This is a lot of work. I know.</div><div><br></div><div><div>Would you say that this entire issue is a part of the general replicability crisis in science? </div></div><div><br></div><div>Danko</div><br clear="all"><div><div dir="ltr" class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr">Dr. Danko Nikolić<br><a href="http://www.danko-nikolic.com" target="_blank">www.danko-nikolic.com</a><br><a href="https://www.linkedin.com/in/danko-nikolic/" target="_blank">https://www.linkedin.com/in/danko-nikolic/</a><div>--- A progress usually starts with an insight ---</div></div></div></div><br></div><div id="DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2"><br>
<table style="border-top:1px solid #d3d4de">
<tr>
<td style="width:55px;padding-top:13px"><a href="https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail" target="_blank"><img src="https://ipmcdn.avast.com/images/icons/icon-envelope-tick-round-orange-animated-no-repeat-v1.gif" alt="" width="46" height="29" style="width: 46px; height: 29px;"></a></td>
<td style="width:470px;padding-top:12px;color:#41424e;font-size:13px;font-family:Arial,Helvetica,sans-serif;line-height:18px">Virus-free. <a href="https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail" target="_blank" style="color:#4453ea">www.avast.com</a>
</td>
</tr>
</table><a href="#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2" width="1" height="1"></a></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Wed, Mar 16, 2022 at 1:00 AM Juyang Weng <<a href="mailto:juyang.weng@gmail.com">juyang.weng@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div dir="ltr">Dear Danko,<div><br></div><div>Since this issue is rampant in AI, I take the liberty to give a CC to connectionist.</div><div>(1) Misleading. Yes, I agree with you. Before I was using fraud and research misconduct, but Nature EIC suggested using a softer term which I agreed to.</div><div>(2) Typo: thank you immensely.</div><div>(3) "p-value in statistical science": I agree with you. My point is to show the necessity. They are not the same.</div><div>(4) "How about other papers in Nature that you did not mention?" I might miss some machine learning papers in Nature since 2015. But almost all machine learning papers since 2015 in Nature and Science suffer from the protocol flaw of Post-Selections Using Test Sets. This protocol flaw is many more machine learning papers that use neural networks, reinforcement learning, swarm mode, reservoir mode, evolutionary mode, and so on. Please suggest what we should do, as much resource is being wasted.</div><div>Best regards,</div><div>-John</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Mar 14, 2022 at 5:54 AM Danko Nikolic <<a href="mailto:danko.nikolic@gmail.com" target="_blank">danko.nikolic@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div>Dear Juyang,</div><div><br></div><div>I just read your post and immediately read the linked paper. I find it interesting. Thank you for that. I am amazed that the extent of the problem is so high.</div><div><br></div><div>I also learned something. I am now completing one study and I realize I did not decouple the random seeds. Thank you for pointing out the need for such decoupling.</div><div><br></div><div>Here some comments from my side:</div><div><br></div><div> - you mention that the results are "misleading", but it seems to me that a more strong term should be used: the results are "too optimistic". The real performances are worse than what has been reported.</div><div><br></div><div> - found a typo: at one place "ransom seed" should be "random seed".</div><div><br></div><div> - a minor comment: You say: " This is similar to, but more transparent than, so-called p-value in statistical science. " I would say that the two things are not too similar. Reporting average + SD is more similar to standard error of the measurement.<br></div><div><br></div><div> Finally, I have a question: How about other papers in Nature that you did not mention? Are there any other papers that have done it right? Should they be mentioned too? Or in other words, do you have any impression of which percentage of papers have done it right? Is it maybe that all of the papers are misleading? Or maybe the split is 50-50?</div><div><br></div><div>I hope you don't mind me sending my thoughts.</div><div><br></div><div>Best,</div><div><br></div><div>Danko</div><div><br></div><div><br></div><div><br></div><div><br></div><div><br></div><br clear="all"><div><div dir="ltr"><div dir="ltr">Dr. Danko Nikolić<br><a href="http://www.danko-nikolic.com" target="_blank">www.danko-nikolic.com</a><br><a href="https://www.linkedin.com/in/danko-nikolic/" target="_blank">https://www.linkedin.com/in/danko-nikolic/</a><div>--- A progress usually starts with an insight ---</div></div></div></div><br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Mar 14, 2022 at 7:53 AM Juyang Weng <<a href="mailto:juyang.weng@gmail.com" target="_blank">juyang.weng@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div>Dear Juergen,</div><div>Your service is appreciated but what you are doing is risky.</div><div>I predict that a large number of them, if not all, are "rising stars" of protocol flaws.</div><div>Please read why:</div><div>------------------------------<br>Message: 2<br>Date: Thu, 10 Mar 2022 17:21:37 -0500<br>From: Juyang Weng <<a href="mailto:juyang.weng@gmail.com" target="_blank">juyang.weng@gmail.com</a>><br>To: Post Connectionists <<a href="mailto:connectionists@mailman.srv.cs.cmu.edu" target="_blank">connectionists@mailman.srv.cs.cmu.edu</a>><br>Subject: Connectionists: A challenge to Post-Selections in Deep<br> Learning<br>Message-ID:<br> <CAJmX=<a href="mailto:6Bx139Ux0iA5PwvEEPzjyXiUcU%2BWQAcUSO9oOQiSCnkxA@mail.gmail.com" target="_blank">6Bx139Ux0iA5PwvEEPzjyXiUcU+WQAcUSO9oOQiSCnkxA@mail.gmail.com</a>><br>Content-Type: text/plain; charset="utf-8"<br><br>Through a review of AI papers published in Nature since 2015, this report<br>discusses the technical flaws called Post-Selection in the charged papers.<br>This report suggests the appropriate protocol, explains reasons for the<br>protocol, why what the papers have done is inappropriate and therefore<br>yields misleading results. The charges below are applicable to whole<br>systems and system components, and in all learning modes, including<br>supervised, reinforcement, and swarm learning modes, since the concepts<br>about training sets, validation sets, and test sets all apply. A<br>reinforcement-learning algorithm includes not only a handcrafted form of<br>task-specific, desired answers but also values of all answers, desired and<br>undesired. A supervised learning method typically does not provide values<br>for intermediate steps (e.g., hidden features), but in contrast, a<br>reinforcement learning mode must provide values for intermediate steps<br>using a greedy search (e.g., time discount). Casting dice is the key<br>protocol flaw that owes a due transparency about all losers (e.g., how good<br>they are). A commercial product is impractical if it requires every<br>customer to cast dice and almost all trained ?lives? must cause accidents<br>and be punished by deaths except the luckiest ?life?. All the losers and<br>the luckiest are unethically determined by so called ?unseen? (in fact<br>should be called ?first seen?) test sets but the human programmer saw all<br>the scores before he decided who are losers and who is the luckiest. Such a<br>deep learning methodology gives no product credibility.<br><br><a href="http://www.cse.msu.edu/~weng/research/2021-06-28-Report-to-Nature-specific-PSUTS.pdf" rel="noreferrer" target="_blank">http://www.cse.msu.edu/%7eweng/research/2021-06-28-Report-to-Nature-specific-PSUTS.pdf</a><br><br></div>---- <div>Date: Fri, 11 Mar 2022 04:57:02 +0000<br>From: Schmidhuber Juergen <<a href="mailto:juergen@idsia.ch" target="_blank">juergen@idsia.ch</a>><br>To: "<a href="mailto:connectionists@cs.cmu.edu" target="_blank">connectionists@cs.cmu.edu</a>" <<a href="mailto:connectionists@cs.cmu.edu" target="_blank">connectionists@cs.cmu.edu</a>><br>Subject: Connectionists: Rising Stars in AI Symposium at KAUST<br>Message-ID: <<a href="mailto:A95E83F0-0E51-451F-B532-BF0FB9DF0CCD@supsi.ch" target="_blank">A95E83F0-0E51-451F-B532-BF0FB9DF0CCD@supsi.ch</a>><br>Content-Type: text/plain; charset="utf-8"<br><br>The AI Initiative at KAUST is hosting the inaugural "Rising Stars in AI Symposium" at KAUST from March 13-15. This event is geared towards young researchers (including Ph.D. students, PostDocs and young faculty), who have recently published promising work at leading AI venues. There will be dozens of brief in-person presentations about papers recently accepted at major AI conferences such as NeurIPS, CVPR, EMNLP, ACL, ICML, ICLR, etc. All speakers will start with an intro for non-AI experts.<br><br>To view the complete program and for more event details, please visit the symposium website:<br><br><a href="https://cemse.kaust.edu.sa/ai/aii-symp-2022" rel="noreferrer" target="_blank">https://cemse.kaust.edu.sa/ai/aii-symp-2022</a><br><br>The symposium will be limited to in-person attendance. So, if you are interested in joining the event, please register here:<br><br><a href="https://docs.google.com/forms/d/e/1FAIpQLSfNnc3N9sGJwkRfePzsZakQSkunhxRadnecGSOd1m7-F7At-A/viewform?hl=en" rel="noreferrer" target="_blank">https://docs.google.com/forms/d/e/1FAIpQLSfNnc3N9sGJwkRfePzsZakQSkunhxRadnecGSOd1m7-F7At-A/viewform?hl=en</a><br><br>We will try our best to accommodate those who register to attend in person.<br><br>J?rgen Schmidhuber<br>Director, AI Initiative, KAUST<br><a href="https://people.idsia.ch/~juergen/kaust-2021.html" rel="noreferrer" target="_blank">https://people.idsia.ch/~juergen/kaust-2021.html</a><br><a href="https://people.idsia.ch/~juergen/kaust-2021-hiring.html" rel="noreferrer" target="_blank">https://people.idsia.ch/~juergen/kaust-2021-hiring.html</a><br clear="all"><div><br></div>-- <br><div dir="ltr"><div dir="ltr">Juyang (John) Weng<br></div></div></div></div>
</blockquote></div>
</blockquote></div><br clear="all"><div><br></div>-- <br><div dir="ltr"><div dir="ltr">Juyang (John) Weng<br></div></div></div>
</blockquote></div>