BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:www.tcs.tifr.res.in/event/128
DTSTAMP:20230914T125911Z
SUMMARY:Stochastic Approximation: Some Theory and an Application
DESCRIPTION:Speaker: Sameer Kamal\nSchool of Technology and Computer Scienc
e\nTata Institute of Fundamental Research\nHomi Bhabha Road\n\nAbstract: \
nWe discuss a vector minmax problem for controlled Markov chains [3]. The
problem of controlling a finite state Markov chain in the presence of an a
dversary so as to ensure desired performance levels for a vector of object
ives is cast in the framework of Blackwell approachability. Relying on an
elementary two time scale construction a control scheme is proposed which
ensures almost sure convergence to the desired set regardless of the adver
sarial actions. This problem serves as an example of stochastic approximat
ion. We conclude with some general theoretical results in stochastic appro
ximation. These relate to stability [2] and sample complexity [1].\n\nRefe
rences\n\n[1] S. Kamal\, On the convergence\, lock-in probability\, and sa
mple complexity of stochastic approximation\, SIAM Journal on Control and
Optimization\, Volume 48\, Number 8\, October 2010\, pp. 5178-5192.\n[2] S
. Kamal\, Stabilization of stochastic approximation by step size adaptatio
n\, Preprint available at http://arxiv.org/abs/1007.4689\n[3] S. Kamal\, A
vector minmax problem for controlled Markov chains\, Preprint available a
t http://arxiv.org/abs/1011.0675\n
URL:https://www.tcs.tifr.res.in/web/events/128
DTSTART;VALUE=DATE:20101129
LOCATION:A-212 (STCS Seminar Room)
END:VEVENT
END:VCALENDAR