BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Neural ODE Processes - Cris Bodnar\, Alex Norcliffe\, Ben Day\, Ja
 cob Moss 
DTSTART:20210216T131500Z
DTEND:20210216T141500Z
UID:TALK155344@talks.cam.ac.uk
CONTACT:Mateja Jamnik
DESCRIPTION:"Join us on Zoom":https://zoom.us/j/99166955895?pwd=SzI0M3pMVE
 kvNmw3Q0dqNDVRalZvdz09\n\nNeural Ordinary Differential Equations (NODEs) u
 se a neural network to model the instantaneous rate of change in the state
  of a system. However\, despite their apparent suitability for dynamics-go
 verned time-series\, NODEs present a few disadvantages. First\, they are u
 nable to adapt to incoming data-points\, a fundamental requirement for rea
 l-time applications imposed by the natural direction of time. Second\, tim
 e-series are often composed of a sparse set of measurements\, which could 
 be explained by many possible underlying dynamics. NODEs do not capture th
 is uncertainty. To this end\, we introduce Neural ODE Processes (NDPs)\, a
  new class of stochastic processes determined by a distribution over Neura
 l ODEs. By maintaining an adaptive data-dependent distribution over the un
 derlying ODE\, we show that our model can successfully capture the dynamic
 s of low-dimensional systems from just a few data-points. At the same time
 \, we demonstrate that NDPs scale up to challenging high-dimensional time-
 series with unknown latent dynamics such as rotating MNIST digits.  
LOCATION:Zoom
END:VEVENT
END:VCALENDAR