<div dir="ltr"><div class="gmail_default" style=""><div class="gmail_default" style="font-size:small;font-family:Arial,Helvetica,sans-serif"><font color="#000000"><span style="letter-spacing:0.2px;white-space:pre-wrap"><font face="georgia, serif"><b>When:    </b>Friday, January 24th, </font></span><b style="font-family:georgia,serif;letter-spacing:0.2px;white-space:pre-wrap"><span style="background-color:rgb(255,255,0)">10:30AM CT</span></b></font></div><div class="gmail_default" style="font-size:small;font-family:Arial,Helvetica,sans-serif"><b style="font-family:georgia,serif;letter-spacing:0.2px;white-space:pre-wrap"><span style="background-color:rgb(255,255,0)"><font color="#000000"><br></font></span></b></div><div class="gmail_default" style="font-size:small;font-family:Arial,Helvetica,sans-serif"><font face="georgia, serif" color="#000000"><b style="letter-spacing:0.2px;white-space:pre-wrap">Where:   </b>Talk will be given<span style="background-color:rgb(255,255,0)"> </span><span style="background-color:rgb(255,255,0)"><font style="font-weight:bold"><u>live, in-person</u></font><font style="font-weight:bold"> </font></span>at</font></div><p class="MsoNormal" style="font-size:small;font-family:Arial,Helvetica,sans-serif;margin:0in;line-height:normal;background-image:initial;background-position:initial;background-size:initial;background-repeat:initial;background-origin:initial;background-clip:initial"><font color="#000000" face="georgia, serif">                   TTIC, 6045 S. Kenwood Avenue</font></p><p class="MsoNormal" style="font-size:small;font-family:Arial,Helvetica,sans-serif;margin:0in;line-height:normal;background-image:initial;background-position:initial;background-size:initial;background-repeat:initial;background-origin:initial;background-clip:initial"><font color="#000000" face="georgia, serif">                    5th Floor, Room 530<b>  </b></font></p><p class="MsoNormal" style="font-size:small;font-family:Arial,Helvetica,sans-serif;margin:0in;line-height:normal;background-image:initial;background-position:initial;background-size:initial;background-repeat:initial;background-origin:initial;background-clip:initial"><font color="#000000" face="georgia, serif"><b><br></b></font></p><div class="gmail_default" style="font-size:small;font-family:Arial,Helvetica,sans-serif"><span style="letter-spacing:0.2px;white-space:pre-wrap"><font face="georgia, serif" color="#000000"><b>Virtually: </b>via Panopto (<a href="https://uchicago.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=96b1e439-8b48-40c3-bd33-afac00e9cf35" target="_blank">Livestream</a>)</font></span></div><div class="gmail_default" style="font-size:small;font-family:Arial,Helvetica,sans-serif"><b style="letter-spacing:0.2px;white-space:pre-wrap"><font face="georgia, serif" color="#000000"><br></font></b></div><div class="gmail_default" style="font-size:small;font-family:Arial,Helvetica,sans-serif"><span style="letter-spacing:0.2px;white-space:pre-wrap"><font face="georgia, serif" color="#000000"><b>Who:    </b> Jamie Simon, UC Berkeley </font></span></div><div class="gmail_default" style="font-size:small;font-family:Arial,Helvetica,sans-serif"><b style="letter-spacing:0.2px;white-space:pre-wrap"><font face="georgia, serif" color="#000000"><br></font></b></div><div class="gmail_default" style="font-size:small"><font face="georgia, serif"><font color="#000000" style=""><b style="letter-spacing:0.2px;white-space:pre-wrap">Title:     </b></font>The Eigenlearning Framework: A Conservation Law Perspective on Kernel Regression and Wide Neural Networks</font><font face="georgia, serif" color="#000000" style=""><font face="Arial, Helvetica, sans-serif"><span style="letter-spacing:0.2px"><br style="letter-spacing:0.2px;white-space:pre-wrap"><br style="letter-spacing:0.2px;white-space:pre-wrap"></span></font><b style="letter-spacing:0.2px;white-space:pre-wrap">Abstract: </b></font><font face="georgia, serif">In a remarkable recent development spanning many works, the generalization of kernel regression -- a classical algorithm equivalent to wide neural nets -- has (in some sense) been solved. We now have a set of simple, closed-form equations which accurately predict things like the expected test risk of the algorithm in terms of the kernel eigenstructure and the target function. In this talk I'll tell this story, with a focus on our own contribution, which is to derive these equations in a simpler manner (no replica calculation or random matrix theory needed, just linear algebra!), give them a nice interpretation in terms of a conserved quantity, and start to apply this powerful theory to knock down problems of interest. Time permitting, I'll talk about how this theory lets us</font></div><font face="georgia, serif" style="font-size:small"><ol><li><font face="georgia, serif" style="font-size:small">explain the "deep bootstrap" phenomenon of Nakkiran et al. (2020),</font></li><li><font face="georgia, serif" style="font-size:small">give a new result on the hardness of the classic parity problem for rotation-invariant kernels,</font></li><li><font face="georgia, serif" style="font-size:small">fashion a theoretical tool for the study of adversarial examples, and</font></li><li><font face="georgia, serif" style="font-size:small">draw a tight analogy between kernel regression and the classic free Fermi gas from statistical physics.</font></li></ol></font><div style="font-size:small;font-family:Arial,Helvetica,sans-serif"><font face="georgia, serif" color="#000000"><br></font></div><div style=""><div class="gmail_default" style=""><font face="georgia, serif" color="#000000" style=""><b style="font-size:small">Bio: </b>Jamie is a</font><font face="georgia, serif" style="font-size:small"> 4th-year PhD student in the physics department at UC Berkeley, aiming to use tools from theoretical physics to build a fundamental understanding of deep neural networks. He is advised by Mike DeWeese. </font></div><font face="georgia, serif" color="#000000" style="font-size:small;font-family:Arial,Helvetica,sans-serif"><br></font><div class="gmail_default" style="font-size:small;font-family:Arial,Helvetica,sans-serif"><font face="georgia, serif" color="#000000">For more information about him and his work, please visit his <a href="https://james-simon.github.io/" target="_blank">personal webpage.</a></font></div></div></div><div><br></div><div><br></div><div><br></div>-- <br><div dir="ltr" class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><b style="background-color:rgb(255,255,255)"><font color="#3d85c6">Brandie Jones </font></b><div><div><div><font color="#3d85c6"><b><i>Executive </i></b></font><b style="color:rgb(61,133,198)"><i>Administrative Assistant</i></b></div></div><div><span style="background-color:rgb(255,255,255)"><font color="#3d85c6">Toyota Technological Institute</font></span></div><div><span style="background-color:rgb(255,255,255)"><font color="#3d85c6">6045 S. Kenwood Avenue</font></span></div><div><span style="background-color:rgb(255,255,255)"><font color="#3d85c6">Chicago, IL  60637</font></span></div></div><div><span style="background-color:rgb(255,255,255)"><font color="#3d85c6"><a href="http://www.ttic.edu" target="_blank">www.ttic.edu</a> </font></span></div><div><font color="#3d85c6"><div style="background-color:rgb(238,238,238)"><div style="color:rgb(32,33,36)"><span style="color:rgb(61,133,198)">Working Remotely on Tuesdays</span><font face="georgia, serif"><br></font></div><div style="color:rgb(32,33,36)"><font color="#3d85c6">Out of Office: Feb 9th - Feb 16th. Remote: Feb 17th - March 3rd.</font></div></div></font></div></div></div></div>