BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:David F. Anderson (University of Wisconsin\, Madison (USA))
DTSTART:20201112T160000Z
DTEND:20201112T163000Z
DTSTAMP:20260421T124007Z
UID:MoRN/1
DESCRIPTION:Title: <a href="https://researchseminars.org/talk/MoRN/1/">Rea
 ction network implementations of neural networks</a>\nby David F. Anderson
  (University of Wisconsin\, Madison (USA)) as part of Seminar on the Mathe
 matics of Reaction Networks\n\n\nAbstract\nI will give an overview of my r
 ecent paper with Badal Joshi and Abhishek Deshpande\, which is entitled "O
 n reaction network implementations of neural networks."  In particular\, I
  will show how reaction networks can be constructed that "implement" a giv
 en neural network.  I will also detail our theoretical results\, which pro
 ve that the ODEs associated with certain reaction network implementations 
 of neural networks have desirable properties including (i) existence of un
 ique positive fixed points that are smooth in the parameters of the model 
 (necessary for gradient descent)\, and (ii) fast convergence to the fixed 
 point regardless of initial condition (necessary for efficient implementat
 ion).  I'll start the talk with a brief primer on neural networks\, but wi
 ll assume familiarity with reaction networks.\n
LOCATION:https://researchseminars.org/talk/MoRN/1/
END:VEVENT
END:VCALENDAR