BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Solving the Bellman equation: Second-order type methods and the du
 al domain approach
DTSTART:20240318T150000
DTEND:20240318T160000
DTSTAMP:20260407T010650Z
UID:18f8b9678f3afb0fe813af5298267722d292c4b22c5672a4296b72cc
CATEGORIES:Conferences - Seminars
DESCRIPTION:Professor Amin Kolarijani\nAbstract: Recent control algorithms
  for Markov decision processes (MDPs) have been designed using an implicit
  analogy with well-established optimization algorithms. In the first part 
 of this talk\, we make this analogy explicit across four problem classes w
 ith a unified solution characterization. This framework\, in turn\, allows
  for a systematic transformation of algorithms from one domain to the othe
 r. In particular\, we identify equivalent optimization and control algorit
 hms that have already been pointed out in the existing literature. With th
 is unifying framework in mind\, we then exploit structural constraints spe
 cific to MDPs for approximating the Hessian for developing second-order-ty
 pe algorithms. In the second part of the talk\, we approach the Bellman eq
 uation through the lens of the conjugate duality. Here\, we show that by e
 xploiting the operational duality of infimal convolution and addition (wit
 h respect to the conjugate transform)\, one can transform the minimization
  in the Bellman equation in the primal domain into an addition in the conj
 ugate domain. In particular\, for a specific class of problems with separa
 ble data in the state and input variables\, this property can be used to r
 educe the typical time complexity of the Bellman operation.\n\n\n\nBio ske
 tch: Amin Kolarijani has a B.Sc. degree in Electrical Engineering from Uni
 versity of Tehran\, Iran\, and an M.Sc. degree in Biomedical Engineering f
 rom Amirkabir University of Technology\, Iran. He received his Ph.D. degre
 e in Systems and Control from Delft University of Technology\, The Netherl
 ands\, in 2022.  He is currently an assistant professor at Delft Center f
 or Systems and Control. His research is mainly focused on developing algor
 ithms for solving dynamic programming and reinforcement learning problems 
 using tools from computational convex analysis\, max-plus algebra\, and al
 gorithmic optimization. He won the Bronze Medal at the Iranian National Ol
 ympiad in Physics in 2005 and was ranked third in Iran’s National Matric
 ulation Exam for Graduate Studies in 2017.\n\n 
LOCATION:ODY 4 03 https://plan.epfl.ch/?room==ODY%204%2003 https://epfl.zo
 om.us/j/67757623931
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR