BEGIN:VCALENDAR

VERSION:2.0

PRODID:-//wp-events-plugin.com//7.4.2//EN

TZID:Asia/Jerusalem

X-WR-TIMEZONE:Asia/Jerusalem
BEGIN:VEVENT

UID:173@math.technion.ac.il

DTSTART;TZID=Asia/Jerusalem:20240221T143000

DTEND;TZID=Asia/Jerusalem:20240221T152000

DTSTAMP:20250109T093656Z

URL:https://math.technion.ac.il/events/idan-mehalel/

SUMMARY:Idan Mehalel (Technion) -- Optimal Prediction Using Expert Advice a
 nd Randomized Littlestone Dimension
DESCRIPTION:Lecturer:Idan Mehalel \n Location:814 Amado\n Suppose that n fo
 recasting experts (or functions) are providing daily rain/no-rain predicti
 ons\, and the best among them is mistaken in at most k many days. For how 
 many days will a person allowed to observe the predictions\, but has no we
 ather-forecasting prior knowledge\, mispredict?\n\nIn this talk\, we will 
 discuss how such classical problems can be reduced to calculating the (ave
 rage) depth of binary trees\, by using newly introduced complexity measure
 s (aka dimensions) of the set of experts/functions. All of those measures 
 are variations of the classical Littlestone dimension (Littlestone ’88).
 \n\nFor the forecasting problem outlined above\, Cesa-Bianchi\, Freund\, H
 elmbold\, and Warmuth [’93\, ’96] provided a nearly optimal bound for 
 deterministic learners\, and left the randomized case as an open problem. 
 We resolve this problem by providing an optimal randomized learner\, and s
 howing that its expected mistake bound equals half of the deterministic bo
 und of Cesa-Bianchi et al.\, up to negligible additive terms.\n\nFor gener
 al (possibly infinite) function classes\, we show that the optimal expecte
 d regret (= #mistakes – k) when learning a function class with Littlesto
 ne dimension d is of order d + (kd)^0.5.\n\nBased on a joint work with Yuv
 al Filmus\, Steve Hanneke\, and Shay Moran.\n 
CATEGORIES:Combinatorics Seminar,Seminars
LOCATION:814 Amado

END:VEVENT

BEGIN:VTIMEZONE

TZID:Asia/Jerusalem

X-LIC-LOCATION:Asia/Jerusalem

BEGIN:STANDARD

DTSTART:20231029T010000

TZOFFSETFROM:+0300

TZOFFSETTO:+0200

TZNAME:IST

END:STANDARD

END:VTIMEZONE
END:VCALENDAR