BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:Kasper Bågmark (Chalmers & GU)
DTSTART:20250512T111500Z
DTEND:20250512T120000Z
DTSTAMP:20260417T003512Z
UID:cam/63
DESCRIPTION:Title: <a href="https://researchseminars.org/talk/cam/63/">Rev
 isiting nonlinear filtering through deep BSDE methods</a>\nby Kasper Bågm
 ark (Chalmers & GU) as part of CAM seminar\n\nLecture held in MV:L14.\n\nA
 bstract\nIn this talk\, I present a new approach to the nonlinear filterin
 g problem based on the deep Backward SDE (BSDE) method. We begin by formul
 ating a system of equations that captures the classical prediction–updat
 e steps in Bayesian filtering. While the update step is tractable up to a 
 normalising constant\, the focus is on approximating the prediction step\,
  which involves evolving the prediction density over time. When the hidden
  state follows a Stochastic Differential Equation (SDE)\, the prediction d
 ensity satisfies the Fokker—Planck equation\, a Partial Differential Equ
 ation (PDE). We solve this PDE using a probabilistic BSDE representation\,
  which we approximate through an optimisation scheme involving neural netw
 orks\, stochastic gradient descent\, and the Euler—Maruyama method. The 
 approach is demonstrated on numerical examples\, and we numerically examin
 e its strong convergence rate.\n
LOCATION:https://researchseminars.org/talk/cam/63/
END:VEVENT
END:VCALENDAR