BEGIN:VCALENDAR VERSION:2.0 PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4// BEGIN:VEVENT UID:20250723T184344EDT-9187mvuRf2@132.216.98.100 DTSTAMP:20250723T224344Z DESCRIPTION:Virtual Informal Systems Seminar (VISS)\, Centre for Intelligen t Machines (CIM) and Groupe d'Etudes et de Recherche en Analyse des Decisi ons (GERAD)\n \n Zoom Link\n Meeting ID: 910 7928 6959        \n Passcode: VIS S\n \n Speaker: Nima Akbarzadeh\, PhD Candidate\, Electrical and Computer En gineering\, 51³Ô¹ÏÍø\n \n Abstract:\n Restless bandits are a class o f sequential resource allocation problems concerned with allocating one or more resources among several alternative processes where the evolution of the process depends on the resource allocated to them.  In 1988\, Whittle proposed an index heuristic for restless bandit problems which has emerge d as a popular solution approach due to its simplicity and strong empirica l performance. The Whittle index heuristic is applicable if the model sati sfies a technical condition known as indexability. We present two general sufficient conditions for indexability and identify simpler to verify refi nements of these conditions for fully-observable models and a class of ind exable models for the partially-observable ones. We show that a generaliza tion of the adaptive greedy algorithm computes the Whittle index for all i ndexable restless bandits. Finally\, we discuss a learning problem in rest less bandits when the dynamics of all processes are unknown initially. The learning algorithm uses a Thompson sampling approach to estimate the unkn own parameters and uses estimated Whittle index to choose actions. We prov ide an upper-bound on the regret with respect to an oracle who knows the t rue parameters.\n \n Bio:\n Nima Akbarzadeh is a Ph.D. candidate in electrica l and computer engineering at 51³Ô¹ÏÍø. He received the B.Sc. deg ree in electrical and computer engineering from Shiraz University\, Iran\, in 2014\, the M.Sc. in electrical and electronics engineering from Bilken t University\, Turkey\, in 2017. He is a recipient of 2020 FRQNT PhD Schol arship. His research interests include stochastic control\, reinforcement learning and multi-armed bandits.\n  \n DTSTART:20210709T141500Z DTEND:20210709T151500Z LOCATION:CA\, ZOOM SUMMARY:Restless Bandits: Indexability\, Whittle Index Computation and Lear ning URL:/cim/channels/event/restless-bandits-indexability- whittle-index-computation-and-learning-331756 END:VEVENT END:VCALENDAR