<div dir="ltr">The points made by Richard, Danko and Tsvi are rock solid to me! I fully second their points!<div>By the way Tsvi, thank you for proposing a novel idea to the community and in fact I went through your paper and video. I was not aware of your work and I found it very stimulating!</div><div><br></div><div>Science needs these debates and more open minded Scientists...</div><div><br></div><div>With many thanks</div><div>Serafim</div><div><br></div><div><br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Sun, 7 Nov 2021 at 00:28, Richard Loosemore <<a href="mailto:rloosemore@susaro.com">rloosemore@susaro.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>
    <div><br>
    </div>
    <div>Adam,</div>
    <div><br>
    </div>
    <div>1) Tsvi Achler has already done the
      things you ask, many times over, so it behooves you to check for
      that before you tell him to do it. Instructing someone to "<span>clearly communicate the novel
        contribution of your approach" when they have already done is is
        an insult.<br>
      </span></div>
    <div><br>
    </div>
    <div>2) The whole point of this discussion
      is that when someone "makes an argument clearly" the community is
      NOT "<span>incredibly open to
        that."  Quite the opposite: the community's attention is fickle,
        tribal, fad-driven, and fundamentally broken.</span></div>
    <div><span><br>
      </span></div>
    <div><span>3) When you say that you "</span><span><span>have trouble believing that
          Google or anyone else will be dismissive of a computational
          approach that actually works," that truly boggles the mind.</span></span></div>
    <div><span><span><br>
        </span></span></div>
    <div><span><span>    a) There is no precise
          definition for "actually works" -- there is no global measure
          of goodness in the space of approaches.</span></span></div>
    <div><span><span><br>
        </span></span></div>
    <div><span><span>    b) Getting the
          attention of someone at e.g. Google is a non-trivial feat in
          itself: just ignoring outsiders is, for Google, a perfectly
          acceptable option.</span></span></div>
    <div><span><span><br>
        </span></span></div>
    <div><span><span>    c) What do you suppose
          would be the reaction of an engineer at Google who gets handed
          a paper by their boss, and is asked "What do you think of
          this?"  Suppose the paper describes an approach that is
          inimicable to what that engineer has been doing their whole
          career. So much so, that if Google goes all-in on this new
          thing, the engineer's skillset will be devalued to junk
          status.  What would the engineer do? They would say "I read
          it. It's just garbage."<br>
        </span></span></div>
    <div><br>
    </div>
    <div>Best</div>
    <div><br>
    </div>
    <div>Richard Loosemore</div>
    <div><br>
    </div>
    <div><br>
    </div>
    <div><br>
    </div>
    <div>On 11/5/21 1:01 PM, Adam Krawitz wrote:<br>
    </div>
    <blockquote type="cite">
      
      
      
      
      <div>
        <p class="MsoNormal"><span lang="EN-US">Tsvi,<u></u><u></u></span></p>
        <p class="MsoNormal"><span><u></u> <u></u></span></p>
        <p class="MsoNormal"><span>I’m
            just a lurker on this list, with no skin in the game, but
            perhaps that gives me a more neutral perspective. In the
            spirit of progress:<u></u><u></u></span></p>
        <p class="MsoNormal"><span><u></u> <u></u></span></p>
        <ol style="margin-top:0cm" type="1" start="1">
          <li style="margin-left:0cm"><span>If you have a neural
              network approach that you feel provides a new and
              important perspective on cognitive processes, then write
              up a paper making that argument clearly, and I think you
              will find that the community is incredibly open to that.
              Yes, if they see holes in the approach they will be
              pointed out, but that is all part of the scientific
              exchange. Examples of this approach include: Elman (1990)
              Finding Structure in Time, Kohonen (1990) The
              Self-Organizing Map, Tenenbaum et al. (2011) How to Grow a
              Mind: Statistics, Structure, and Abstraction (not neural
              nets, but a “new” approach to modelling cognition). I’m
              sure others can provide more examples.<u></u><u></u></span></li>
          <li style="margin-left:0cm"><span>I’m much less familiar
              with how things work on the applied side, but I have
              trouble believing that Google or anyone else will be
              dismissive of a computational approach that actually
              works. Why would they? They just want to solve problems
              efficiently. Demonstrate that your approach can solve a
              problem more effectively (or at least as effectively) as
              the existing approaches, and they will come running.
              Examples of this include: Tesauro’s TD-Gammon, which was
              influential in demonstrating the power of RL, and LeCun et
              al.’s convolutional NN for the MNIST digits.<u></u><u></u></span></li>
        </ol>
        <p class="MsoNormal"><span><u></u> <u></u></span></p>
        <p class="MsoNormal"><span>Clearly
            communicate the novel contribution of your approach and I
            think you will find a receptive audience.<u></u><u></u></span></p>
        <p class="MsoNormal"><span><u></u> <u></u></span></p>
        <p class="MsoNormal"><span>Thanks,<u></u><u></u></span></p>
        <p class="MsoNormal"><span>Adam<u></u><u></u></span></p>
        <p class="MsoNormal"><span><u></u> <u></u></span></p>
        <p class="MsoNormal"><span><u></u> <u></u></span></p>
        <div style="border-right:none;border-bottom:none;border-left:none;border-top:1pt solid rgb(225,225,225);padding:3pt 0cm 0cm">
          <p class="MsoNormal"><b><span lang="EN-US">From:</span></b><span lang="EN-US"> Connectionists
              <a href="mailto:connectionists-bounces@mailman.srv.cs.cmu.edu" target="_blank"><connectionists-bounces@mailman.srv.cs.cmu.edu></a>
              <b>On Behalf Of </b>Tsvi Achler<br>
              <b>Sent:</b> November 4, 2021 9:46 AM<br>
              <b>To:</b> <a href="mailto:gary@ucsd.edu" target="_blank">gary@ucsd.edu</a><br>
              <b>Cc:</b> <a href="mailto:connectionists@cs.cmu.edu" target="_blank">connectionists@cs.cmu.edu</a><br>
              <b>Subject:</b> Re: Connectionists: Scientific Integrity,
              the 2021 Turing Lecture, etc.<u></u><u></u></span></p>
        </div>
        <p class="MsoNormal"><u></u> <u></u></p>
        <div>
          <div>
            <p class="MsoNormal">Lastly Feedforward methods are
              predominant in a large part because they have financial
              backing from large companies with advertising and clout
              like Google and the self-driving craze that never fully
              materialized.  <u></u><u></u></p>
          </div>
          <div>
            <p class="MsoNormal"><u></u> <u></u></p>
          </div>
          <div>
            <p class="MsoNormal">Feedforward methods are not fully
              connectionist unless rehearsal for learning is implemented
              with neurons.  That means storing all patterns, mixing
              them randomly and then presenting to a network to learn. 
              As far as I know, no one is doing this in the community,
              so feedforward methods are only partially connectionist. 
              By allowing popularity to predominate and choking off
              funds and presentation of alternatives we are cheating
              ourselves from pursuing other more rigorous brain-like
              methods.<u></u><u></u></p>
          </div>
          <div>
            <p class="MsoNormal"><u></u> <u></u></p>
          </div>
          <div>
            <p class="MsoNormal">Sincerely,<u></u><u></u></p>
          </div>
          <div>
            <p class="MsoNormal">-Tsvi<u></u><u></u></p>
          </div>
          <div>
            <p class="MsoNormal"><u></u> <u></u></p>
          </div>
        </div>
        <p class="MsoNormal"><u></u> <u></u></p>
        <div>
          <p class="MsoNormal">On Tue, Nov 2, 2021 at 7:08 PM Tsvi
            Achler <<a href="mailto:achler@gmail.com" target="_blank">achler@gmail.com</a>>
            wrote:<u></u><u></u></p>
        </div>
        <p class="MsoNormal">Gary- Thanks for the accessible online link
          to the book. <u></u><u></u></p>
        <div>
          <p class="MsoNormal"><u></u> <u></u></p>
        </div>
        <div>
          <p class="MsoNormal">I looked especially at the inhibitory
            feedback section of the book which describes an Air
            Conditioner AC type feedback.  <u></u><u></u></p>
        </div>
        <div>
          <p class="MsoNormal">It then describes a general field-like
            inhibition based on all activations in the layer.  It also
            describes the role of inhibition in sparsity and feedforward
            inhibition,<u></u><u></u></p>
        </div>
        <div>
          <p class="MsoNormal"><u></u> <u></u></p>
        </div>
        <div>
          <p class="MsoNormal">The feedback described in Regulatory
            Feedback is similar to the AC feedback but occurs for each
            neuron individually, vis-a-vis its inputs.<u></u><u></u></p>
        </div>
        <div>
          <p class="MsoNormal">Thus for context, regulatory feedback is
            not a field-like inhibition, it is very directed based on
            the neurons that are activated and their inputs.  This sort
            of regulation is also the foundation of Homeostatic
            Plasticity findings (albeit with changes in Homeostatic
            regulation in experiments occurring in a slower time
            scale).  The regulatory feedback model describes the effect
            and role in recognition of those regulated connections in
            real time during recognition.<u></u><u></u></p>
        </div>
        <div>
          <p class="MsoNormal"><u></u> <u></u></p>
        </div>
        <div>
          <p class="MsoNormal">I would be happy to discuss further and
            collaborate on writing about the differences between the
            approaches for the next book or review.<u></u><u></u></p>
        </div>
        <div>
          <p class="MsoNormal"><u></u> <u></u></p>
        </div>
        <div>
          <p class="MsoNormal">And I want to point out to folks, that
            the system is based on politics and that is why certain work
            is not cited like it should, but even worse these politics
            are here in the group today and they continue to very
            strongly influence decisions in the connectionist community
            and holds us back.<u></u><u></u></p>
        </div>
        <div>
          <p class="MsoNormal"><u></u> <u></u></p>
        </div>
        <div>
          <p class="MsoNormal">Sincerely,<u></u><u></u></p>
        </div>
        <div>
          <p class="MsoNormal">-Tsvi<u></u><u></u></p>
        </div>
        <p class="MsoNormal"><u></u> <u></u></p>
        <div>
          <p class="MsoNormal">On Mon, Nov 1, 2021 at 10:59 AM <a href="mailto:gary@ucsd.edu" target="_blank">
              gary@ucsd.edu</a> <<a href="mailto:gary@eng.ucsd.edu" target="_blank">gary@eng.ucsd.edu</a>>
            wrote:<u></u><u></u></p>
        </div>
        <div>
          <div>
            <p class="MsoNormal"><span>Tsvi - While I think
                <a href="https://www.amazon.com/dp/0262650541/" target="_blank">Randy and
                  Yuko's book
                </a>is actually somewhat better than the online version
                (and buying choices on amazon start at $9.99), there
                <b>is</b> <a href="https://compcogneuro.org/" target="_blank">an online
                  version.</a> <u></u><u></u></span></p>
          </div>
          <div>
            <p class="MsoNormal"><span>Randy & Yuko's models take into
                account feedback and inhibition. <u></u><u></u></span></p>
          </div>
        </div>
        <p class="MsoNormal"><u></u> <u></u></p>
        <div>
          <div>
            <p class="MsoNormal">On Mon, Nov 1, 2021 at 10:05 AM Tsvi
              Achler <<a href="mailto:achler@gmail.com" target="_blank">achler@gmail.com</a>>
              wrote:<u></u><u></u></p>
          </div>
          <blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin-left:4.8pt;margin-right:0cm">
            <div>
              <div>
                <p class="MsoNormal">Daniel,<u></u><u></u></p>
                <div>
                  <p class="MsoNormal"><u></u> <u></u></p>
                </div>
                <div>
                  <p class="MsoNormal">Does your book include a
                    discussion of Regulatory or Inhibitory Feedback
                    published in several low impact journals between
                    2008 and 2014 (and in videos subsequently)?<u></u><u></u></p>
                </div>
                <div>
                  <p class="MsoNormal">These are networks where the
                    primary computation is inhibition back to the inputs
                    that activated them and may be very counterintuitive
                    given today's trends.  You can almost think of them
                    as the opposite of Hopfield networks.<u></u><u></u></p>
                </div>
                <div>
                  <p class="MsoNormal"><u></u> <u></u></p>
                </div>
                <div>
                  <p class="MsoNormal">I would love to check inside the
                    book but I dont have an academic budget that allows
                    me access to it and that is a huge part of the
                    problem with how information is shared and
                    funding is allocated. I could not get access to any
                    of the text or citations especially Chapter 4:
                    "Competition, Lateral Inhibition, and Short-Term
                    Memory", to weigh in.<u></u><u></u></p>
                </div>
                <div>
                  <p class="MsoNormal"><u></u> <u></u></p>
                </div>
                <div>
                  <p class="MsoNormal">I wish the best circulation for
                    your book, but even if the Regulatory Feedback Model
                    is in the book, that does not change the fundamental
                    problem if the book is not readily available. <u></u><u></u></p>
                </div>
                <div>
                  <p class="MsoNormal"><u></u> <u></u></p>
                </div>
                <div>
                  <p class="MsoNormal">The same goes with Steve
                    Grossberg's book, I cannot easily look inside.  With
                    regards to Adaptive Resonance I dont subscribe to
                    lateral inhibition as a predominant mechanism, but I
                    do believe a function such as vigilance is very
                    important during recognition and Adaptive Resonance
                    is one of a very few models that have it.  The
                    Regulatory Feedback model I have developed (and
                    Michael Spratling studies a similar model as well)
                    is built primarily using the vigilance type of
                    connections and allows multiple neurons to be
                    evaluated at the same time and continuously during
                    recognition in order to determine which (single or
                    multiple neurons together) match the inputs the best
                    without lateral inhibition.<u></u><u></u></p>
                </div>
                <div>
                  <p class="MsoNormal"><u></u> <u></u></p>
                </div>
                <div>
                  <p class="MsoNormal">Unfortunately within conferences
                    and talks predominated by the Adaptive Resonance
                    crowd I have experienced the familiar dismissiveness
                    and did not have an opportunity to give a proper
                    talk. This goes back to the larger issue of academic
                    politics based on small self-selected committees,
                    the same issues that exist with the feedforward
                    crowd, and pretty much all of academia.<u></u><u></u></p>
                </div>
                <div>
                  <p class="MsoNormal"><u></u> <u></u></p>
                </div>
                <div>
                  <p class="MsoNormal">Today's information age
                    algorithms such as Google's can determine relevance
                    of information and ways to display them, but
                    hegemony of the journal systems and the small
                    committee system of academia developed in the middle
                    ages (and their mutual synergies) block the use of
                    more modern methods in research.  Thus we are stuck
                    with this problem, which especially affects those
                    that are trying to introduce something new and
                    counterintuitive, and hence the results described in
                    the two National Bureau of Economic Research
                    articles I cited in my previous message.<u></u><u></u></p>
                </div>
                <div>
                  <p class="MsoNormal"><u></u> <u></u></p>
                </div>
                <div>
                  <p class="MsoNormal"><span style="color:black">Thomas,
                      I am happy to have more discussions and/or start a
                      different thread.</span><u></u><u></u></p>
                </div>
                <div>
                  <p class="MsoNormal"><u></u> <u></u></p>
                </div>
                <div>
                  <p class="MsoNormal">Sincerely,<u></u><u></u></p>
                </div>
                <div>
                  <p class="MsoNormal">Tsvi Achler MD/PhD<u></u><u></u></p>
                </div>
                <div>
                  <p class="MsoNormal"><u></u> <u></u></p>
                </div>
                <div>
                  <p class="MsoNormal"><u></u> <u></u></p>
                </div>
              </div>
              <p class="MsoNormal"><u></u> <u></u></p>
              <div>
                <div>
                  <p class="MsoNormal">On Sun, Oct 31, 2021 at 12:49 PM
                    Levine, Daniel S <<a href="mailto:levine@uta.edu" target="_blank">levine@uta.edu</a>>
                    wrote:<u></u><u></u></p>
                </div>
                <blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin-left:4.8pt;margin-right:0cm">
                  <div>
                    <div>
                      <p class="MsoNormal"><span style="font-size:12pt;color:black">Tsvi,<u></u><u></u></span></p>
                    </div>
                    <div>
                      <p class="MsoNormal"><span style="font-size:12pt;color:black"><u></u> <u></u></span></p>
                    </div>
                    <div>
                      <p class="MsoNormal"><span style="font-size:12pt;color:black">While
                          deep learning and feedforward networks have an
                          outsize popularity, there are plenty of
                          published sources that cover a much wider
                          variety of networks, many of them more
                          biologically based than deep learning.  A
                          treatment of a range of neural network
                          approaches, going from simpler to more complex
                          cognitive functions, is found in my textbook
                          <i>Introduction to Neural and Cognitive
                            Modeling</i> (3rd edition, Routledge,
                          2019).  Also Steve Grossberg's book
                          <i>Conscious Mind, Resonant Brain</i> (Oxford,
                          2021) emphasizes a variety of architectures
                          with a strong biological basis.<u></u><u></u></span></p>
                    </div>
                    <div>
                      <p class="MsoNormal"><span style="font-size:12pt;color:black"><u></u> <u></u></span></p>
                    </div>
                    <div>
                      <p class="MsoNormal"><span style="font-size:12pt;color:black"><u></u> <u></u></span></p>
                    </div>
                    <div>
                      <p class="MsoNormal"><span style="font-size:12pt;color:black">Best,<u></u><u></u></span></p>
                    </div>
                    <div>
                      <p class="MsoNormal"><span style="font-size:12pt;color:black"><u></u> <u></u></span></p>
                    </div>
                    <div>
                      <p class="MsoNormal"><span style="font-size:12pt;color:black"><u></u> <u></u></span></p>
                    </div>
                    <div>
                      <p class="MsoNormal"><span style="font-size:12pt;color:black">Dan
                          Levine<u></u><u></u></span></p>
                    </div>
                    <div class="MsoNormal" style="text-align:center" align="center">
                      <hr width="98%" size="2" align="center">
                    </div>
                    <div id="gmail-m_-1845407307866719376gmail-m_-3449454411413463923gmail-m_-99491911606122447gmail-m_-1743973086040584978m_-7977962918574028451m_3158366256145482305m_-7622355768744816385m_-898643676068276388gmail-m_-7809501330019106925gmail-m_8996447038276730094gmail-m_-2305817410909496922gmail-m_7665975300539281535divRplyFwdMsg">
                      <p class="MsoNormal"><b><span style="color:black">From:</span></b><span style="color:black"> Connectionists <<a href="mailto:connectionists-bounces@mailman.srv.cs.cmu.edu" target="_blank">connectionists-bounces@mailman.srv.cs.cmu.edu</a>>
                          on behalf of Tsvi Achler <<a href="mailto:achler@gmail.com" target="_blank">achler@gmail.com</a>><br>
                          <b>Sent:</b> Saturday, October 30, 2021 3:13
                          AM<br>
                          <b>To:</b> Schmidhuber Juergen <<a href="mailto:juergen@idsia.ch" target="_blank">juergen@idsia.ch</a>><br>
                          <b>Cc:</b> <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>
                          <b>Subject:</b> Re: Connectionists: Scientific
                          Integrity, the 2021 Turing Lecture, etc.</span>
                        <u></u><u></u></p>
                      <div>
                        <p class="MsoNormal"> <u></u><u></u></p>
                      </div>
                    </div>
                    <div>
                      <div>
                        <p class="MsoNormal">Since the title of the
                          thread is Scientific Integrity, I want to
                          point out some issues about trends in academia
                          and then especially focusing on the
                          connectionist community.
                          <u></u><u></u></p>
                        <div>
                          <p class="MsoNormal"><u></u> <u></u></p>
                          <div>
                            <p class="MsoNormal">In general analyzing
                              impact factors etc the most important
                              progress gets silenced until the
                              mainstream picks it up <a href="https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.nber.org%2Fsystem%2Ffiles%2Fworking_papers%2Fw22180%2Fw22180.pdf%3Ffbclid%3DIwAR1zHhU4wmkrHASTaE-6zwIs6gI9-FxZcCED3BETxUJlMsbN_2hNbmJAmOA&data=04%7C01%7Clevine%40uta.edu%7Cb1a267e3b6a64ada666208d99ca37f6d%7C5cdc5b43d7be4caa8173729e3b0a62d9%7C1%7C0%7C637713048300122043%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=9o%2FzcYY8gZVZiAwyEL5SVI9TEzBWfKf7nfhdWWg8LHU%3D&reserved=0" target="_blank">Impact
                                Factiors in novel research
                                www.nber.org/.../working_papers/w22180/w22180.pdf</a>  and
                              often this may take a generation <a href="https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.nber.org%2Fdigest%2Fmar16%2Fdoes-science-advance-one-funeral-time%3Ffbclid%3DIwAR1Lodsf1bzje-yQU9DvoZE2__S6R7UPEgY1_LxZCSLdoAYnj-uco0JuyVk&data=04%7C01%7Clevine%40uta.edu%7Cb1a267e3b6a64ada666208d99ca37f6d%7C5cdc5b43d7be4caa8173729e3b0a62d9%7C1%7C0%7C637713048300132034%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=DgxnJTT7MsN5KCzZlA7VAHKrHXVsRsYhopJv0FCwbtw%3D&reserved=0" target="_blank">https://www.nber.org/.../does-science-advance-one-funeral...</a>  .<u></u><u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal"><u></u> <u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal">The connectionist field
                              is stuck on feedforward networks and
                              variants such as with inhibition of
                              competitors (e.g. lateral inhibition), or
                              other variants that are sometimes labeled
                              as recurrent networks for learning time
                              where the feedforward networks can be
                              rewound in time.<u></u><u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal"><u></u> <u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal">This stasis is
                              specifically occuring with the popularity
                              of deep learning.  This is often portrayed
                              as neurally plausible connectionism but
                              requires an implausible amount of
                              rehearsal and is not connectionist if this
                              rehearsal is not implemented with neurons
                              (see video link for further
                              clarification).<u></u><u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal"><u></u> <u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal">Models which have true
                              feedback (e.g. back to their own inputs)
                              cannot learn by backpropagation but there
                              is plenty of evidence these types of
                              connections exist in the brain and are
                              used during recognition. Thus they get
                              ignored: no talks in universities, no
                              featuring in "premier" journals and no
                              funding. <u></u><u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal"><u></u> <u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal">But they are important
                              and may negate the need for rehearsal as
                              needed in feedforward methods.  Thus may
                              be essential for moving connectionism
                              forward.<u></u><u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal"><u></u> <u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal">If the community is
                              truly dedicated to brain motivated
                              algorithms, I recommend giving more time
                              to networks other than feedforward
                              networks.<u></u><u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal"><u></u> <u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal">Video: <a href="https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3Dm2qee6j5eew%26list%3DPL4nMP8F3B7bg3cNWWwLG8BX-wER2PeB-3%26index%3D2&data=04%7C01%7Clevine%40uta.edu%7Cb1a267e3b6a64ada666208d99ca37f6d%7C5cdc5b43d7be4caa8173729e3b0a62d9%7C1%7C0%7C637713048300132034%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=EaEp5zLZ7HkDhsBHmP3x3ObPl8j14B8%2BFcOkkNEWZ9w%3D&reserved=0" target="_blank">https://www.youtube.com/watch?v=m2qee6j5eew&list=PL4nMP8F3B7bg3cNWWwLG8BX-wER2PeB-3&index=2</a><u></u><u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal"><u></u> <u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal">Sincerely,<u></u><u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal">Tsvi Achler<u></u><u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal"><u></u> <u></u></p>
                          </div>
                          <div>
                            <p class="MsoNormal"><u></u> <u></u></p>
                          </div>
                        </div>
                      </div>
                      <p class="MsoNormal"><u></u> <u></u></p>
                      <div>
                        <div>
                          <p class="MsoNormal">On Wed, Oct 27, 2021 at
                            2:24 AM Schmidhuber Juergen <<a href="mailto:juergen@idsia.ch" target="_blank">juergen@idsia.ch</a>>
                            wrote:<u></u><u></u></p>
                        </div>
                        <blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0cm 0cm 0cm 6pt;margin-left:4.8pt;margin-right:0cm">
                          <p class="MsoNormal" style="margin-bottom:12pt">Hi, fellow
                            artificial neural network enthusiasts!<br>
                            <br>
                            The connectionists mailing list is perhaps
                            the oldest mailing list on ANNs, and many
                            neural net pioneers are still subscribed to
                            it. I am hoping that some of them - as well
                            as their contemporaries - might be able to
                            provide additional valuable insights into
                            the history of the field.<br>
                            <br>
                            Following the great success of massive open
                            online peer review (MOOR) for my 2015 survey
                            of deep learning (now the most cited article
                            ever published in the journal Neural
                            Networks), I've decided to put forward
                            another piece for MOOR. I want to thank the
                            many experts who have already provided me
                            with comments on it. Please send additional
                            relevant references and suggestions for
                            improvements for the following draft
                            directly to me at
                            <a href="mailto:juergen@idsia.ch" target="_blank">juergen@idsia.ch</a>:<br>
                            <br>
                            <a href="https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpeople.idsia.ch%2F~juergen%2Fscientific-integrity-turing-award-deep-learning.html&data=04%7C01%7Clevine%40uta.edu%7Cb1a267e3b6a64ada666208d99ca37f6d%7C5cdc5b43d7be4caa8173729e3b0a62d9%7C1%7C0%7C637713048300142030%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=mW3lH7SqKg4EuJfDwKcC2VhwEloC3ndh6kI5gfQ2Ofw%3D&reserved=0" target="_blank">https://people.idsia.ch/~juergen/scientific-integrity-turing-award-deep-learning.html</a><br>
                            <br>
                            The above is a point-for-point critique of
                            factual errors in ACM's justification of the
                            ACM A. M. Turing Award for deep learning and
                            a critique of the Turing Lecture published
                            by ACM in July 2021. This work can also be
                            seen as a short history of deep learning, at
                            least as far as ACM's errors and the Turing
                            Lecture are concerned.<br>
                            <br>
                            I know that some view this as a
                            controversial topic. However, it is the very
                            nature of science to resolve controversies
                            through facts. Credit assignment is as core
                            to scientific history as it is to machine
                            learning. My aim is to ensure that the true
                            history of our field is preserved for
                            posterity.<br>
                            <br>
                            Thank you all in advance for your help! <br>
                            <br>
                            Jürgen Schmidhuber<br>
                            <br>
                            <br>
                            <br>
                            <br>
                            <br>
                            <br>
                            <br>
                            <u></u><u></u></p>
                        </blockquote>
                      </div>
                    </div>
                  </div>
                </blockquote>
              </div>
            </div>
          </blockquote>
        </div>
        <p class="MsoNormal"><br clear="all">
          <u></u><u></u></p>
        <div>
          <p class="MsoNormal"><u></u> <u></u></p>
        </div>
        <p class="MsoNormal">-- <u></u><u></u></p>
        <div>
          <p class="MsoNormal">Gary Cottrell 858-534-6640 FAX:
            858-534-7029<u></u><u></u></p>
        </div>
        <div>
          <p class="MsoNormal" style="margin-bottom:12pt">Computer
            Science and Engineering 0404<br>
            IF USING FEDEX INCLUDE THE FOLLOWING LINE:    <br>
            CSE Building, Room 4130<br>
            University of California San Diego                          
                       -<br>
            9500 Gilman Drive # 0404<br>
            La Jolla, Ca. 92093-0404<u></u><u></u></p>
        </div>
        <div>
          <p class="MsoNormal">Email: <a href="mailto:gary@ucsd.edu" target="_blank">gary@ucsd.edu</a><br>
            Home page: <a href="http://www-cse.ucsd.edu/~gary/" target="_blank">http://www-cse.ucsd.edu/~gary/</a><u></u><u></u></p>
        </div>
        <div>
          <p class="MsoNormal"><span style="font-size:9.5pt">Schedule: <a href="http://tinyurl.com/b7gxpwo" target="_blank">http://tinyurl.com/b7gxpwo</a></span><u></u><u></u></p>
        </div>
        <div>
          <p class="MsoNormal"><span style="font-size:9.5pt"><u></u> <u></u></span></p>
        </div>
        <p><i><span>Listen carefully,</span></i><i><span><br>
            </span></i><i><span>Neither the Vedas</span></i><i><span><br>
            </span></i><i><span>Nor the Qur'an</span></i><i><span><br>
            </span></i><i><span>Will teach you this:</span></i><i><span><br>
            </span></i><i><span>Put the bit in its mouth,</span></i><i><span><br>
            </span></i><i><span>The saddle on its back,</span></i><i><span><br>
            </span></i><i><span>Your foot in the stirrup,</span></i><i><span><br>
            </span></i><i><span>And ride your wild
              runaway mind</span></i><i><span><br>
            </span></i><i><span>All the way to heaven.<u></u><u></u></span></i></p>
        <p><i><span>-- Kabir</span></i><i><span><u></u><u></u></span></i></p>
      </div>
    </blockquote>
    <p><br>
    </p>
  </div>
</blockquote></div><br clear="all"><div><br></div>-- <br><div dir="ltr" class="gmail_signature">          
<div>
<span style="border-collapse:separate;color:rgb(0,0,0);font-family:Helvetica;font-style:normal;font-variant:normal;font-weight:normal;letter-spacing:normal;line-height:normal;text-align:-webkit-auto;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;font-size:medium"><span style="border-collapse:separate;color:rgb(0,0,0);font-family:Helvetica;font-style:normal;font-variant:normal;font-weight:normal;letter-spacing:normal;line-height:normal;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;font-size:medium"><div style="overflow-wrap: break-word;"><span><span style="font-size:12px"><div style="font-size:12px"><span style="font-size:15px;font-family:Calibri">Serafim Rodrigues</span></div><div style="overflow-wrap: break-word;"><font face="Calibri" size="4"><span style="font-size:15px"><span style="font-family:Helvetica;font-size:12px"><div style="overflow-wrap: break-word;"><span style="font-family:Calibri;font-size:15px"><font color="#31557F">Group Leader<br></font><b>BCAM - <font color="#284FAE"></font></b><font color="#1769FD"><font color="#1C48E9">Basque Center for Applied Mathematics</font><br></font>Alameda de Mazarredo, 14<br>E-48009 Bilbao, Basque Country - Spain<br>Tel. +34 946 567 842</span></div><div style="overflow-wrap: break-word;"><font face="Calibri" size="4"><span style="font-size:15px"><a href="mailto:srodrigues@bcamath.org" target="_blank">srodrigues@bcamath.org</a> | <a href="http://www.bcamath.org/srodrigues" target="_blank">www.bcamath.org/srodrigues</a></span></font></div><div style="overflow-wrap: break-word;"><font face="Calibri" color="#7F7F7F" size="4"><span style="font-size:15px"><b><i><br></i></b></span></font></div><div style="overflow-wrap: break-word;"><span style="font-family:Calibri;font-size:15px"><font color="#7F7F7F"><b><i>(</i></b><i></i></font><i><font color="#0041F1">matematika</font><font color="#7F7F7F"> mugaz bestalde<b>)</b></font></i></span></div></span></span></font></div></span></span></div></span>
</span>
</div>

</div>