BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Date iCal//NONSGML kigkonsult.se iCalcreator 2.20.2//
METHOD:PUBLISH
X-WR-CALNAME;VALUE=TEXT:Ical
BEGIN:VTIMEZONE
TZID:Asia/Jerusalem
BEGIN:STANDARD
DTSTART:20151025T020000
TZOFFSETFROM:+0300
TZOFFSETTO:+0200
TZNAME:IST
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20160325T020000
TZOFFSETFROM:+0200
TZOFFSETTO:+0300
TZNAME:IDT
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.50782.field_date.0@mathematics.huji.ac.il
DTSTAMP:20210308T092236Z
CREATED:20180114T000005Z
DESCRIPTION:Date: \n\n3:30pm to 4:30pm\n\n\nSee also: Game Theory & Mathe
matical Economics\, Events & Seminars\, SeminarsLocation: \n\nElath Hall\,
2nd floor\, Feldman Building\, Edmond Safra Campus\n\n\nTopic: Calibrate
d Forecasts\, Leaks\, and Game Equilibria (joint work with Dean P. Foster)
\nPlace: Elath Hall\, 2nd floor\, Feldman Building\, Edmond Safra Campus
\nTime: Sunday\, November 22\, 2015 at 4:00 p.m.\n Refres
hments available at 3:30 p.m.\n YOU ARE CORDIALLY
INVITED\nAbstract: How good is a forecaster? Assume for concreteness that
every day the forecaster issues a forecast of the type 'the chance of rain
tomorrow is 30%.' A simple test one may conduct is to calculate the propo
rtion of rainy days out of those days that the forecast was 30%\, and comp
are it to 30%\; and do the same for all other forecasts. A forecaster is s
aid to be _calibrated_ if\, in the long run\, the differences between the
actual proportions of rainy days and the forecasts are small—no matter wha
t the weather really was. We start from the classical result that calibrat
ion can always be guaranteed by _randomized_ forecasting procedures. We th
en _smooth out_ the calibration score by combining nearby forecasts. While
regular calibration can be guaranteed only by randomized forecasting proc
edures\, smooth calibration can be guaranteed by _deterministic_ procedure
s. As a consequence\, it does not matter if the forecasts are _leaked_\, i
.e.\, made known in advance: smooth calibration can nevertheless be guaran
teed (while regular calibration cannot). Moreover\, our procedure has fini
te recall\, is stationary\, and all forecasts lie on a finite grid. We als
o consider related problems: online linear regression\, weak calibration\,
and _uncoupled Nash dynamics_ in n-person games (vs. correlated equilibri
a dynamics for calibrated learning).
DTSTART;TZID=Asia/Jerusalem:20151122T153000
DTEND;TZID=Asia/Jerusalem:20151122T163000
LAST-MODIFIED:20191104T000004Z
SUMMARY:Game theory & economics: Sergiu Hart (HUJI)\, 'Calibrated Forecasts
\, Leaks\, and Game Equilibria (joint work with Dean P. Foster)'
URL;TYPE=URI:https://mathematics.huji.ac.il/event/game-theory-economics-ser
giu-hart-huji-calibrated-forecasts-leaks-and-game-equilibria
END:VEVENT
END:VCALENDAR