BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Ashoka University - ECPv6.17.4//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:Ashoka University
X-ORIGINAL-URL:https://www.ashoka.edu.in
X-WR-CALDESC:Events for Ashoka University
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:Asia/Kolkata
BEGIN:STANDARD
TZOFFSETFROM:+0530
TZOFFSETTO:+0530
TZNAME:IST
DTSTART:20230101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20240919T183000
DTEND;TZID=Asia/Kolkata:20241125T193000
DTSTAMP:20240829T035302Z
CREATED:20240829T035302Z
LAST-MODIFIED:20240829T035302Z
UID:61866-1726770600-1732563000@www.ashoka.edu.in
SUMMARY:SOA Colloquium Series
DESCRIPTION:
URL:https://www.ashoka.edu.in/event/soa-colloquium-series/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20241015T133000
DTEND;TZID=Asia/Kolkata:20241015T143000
DTSTAMP:20241009T001009Z
CREATED:20241009T001009Z
LAST-MODIFIED:20241009T001009Z
UID:64567-1728999000-1729002600@www.ashoka.edu.in
SUMMARY:CS Colloquium
DESCRIPTION:Date: 15-October-2024\nTime: 13:30 to 14:30\nVenue: AC04-LR-009 \nTitle: Optimal Top-Two Method for Best Arm Identification and Fluid Analysis \nAbstract: Top-2 methods are widely used to solve the best arm identification (BAI) problem\, where the goal is to find the arm with the highest mean from a finite set of options. In these methods\, an algorithm pulls the empirical best arm with a fixed probability\, denoted by β\, and pulls the next-best challenger arm otherwise. The algorithm guarantees that the probability of incorrect selection stays below a specified threshold\, δ. Information-theoretic lower bounds on the number of samples needed for the BAI problem are well established and are asymptotically met by plug-in methods\, which are computationally intensive as δ approaches zero. The top-2 algorithm\, for any β between 0 and 1\, achieves a sample complexity within a constant factor of this lower bound. However\, determining the optimal β that exactly matches the lower bound has been a challenge. In this paper\, we tackle this issue by proposing an optimal top-2 algorithm. We use a function of allocations that is anchored to a threshold—if the function exceeds this threshold\, the algorithm pulls the empirical best arm; otherwise\, it pulls the challenger arm. We prove that this algorithm is optimal as δ approaches zero. Our analysis involves studying the limiting behavior of the allocations\, which can be described by a set of ordinary differential equations (ODEs) that are piecewise connected. Using the implicit function theorem\, we show the existence and uniqueness of these ODEs\, and we demonstrate that our proposed algorithm closely follows their solution. This work is based on a collaboration with Prof. Sandeep Juneja (Ashoka University) and Dr. Shubhada Agrawal (Carnegie Mellon University). \nAbout the Speaker: Agniv is a PhD student at the School of Technology and Computer Science\, Tata Institute of Fundamental Research (TIFR) Mumbai. He holds a BTech degree in Mathematics and Computing from the Indian Institute of Technology (IIT) Guwahati. Agniv's research focuses on sequential learning and applied probability.\n  \nYou can also opt-in to the mailing list and calendar notifications: \nCS Colloquium Opt-in Mailing List \nCS Colloquium Calendar
URL:https://www.ashoka.edu.in/event/cs-colloquium/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20241015T134000
DTEND;TZID=Asia/Kolkata:20241015T145000
DTSTAMP:20241014T020207Z
CREATED:20241014T020207Z
LAST-MODIFIED:20241014T020207Z
UID:65029-1728999600-1729003800@www.ashoka.edu.in
SUMMARY:Special Seminar by Prof. Laurence M. Ball
DESCRIPTION:Dear All\, \nThe Department of Economics & ISAAC Centre for Public Policy invites you for the Special Seminar on Tuesday\, October 15th\, 2024. \nTime: 1:40 PM to 2:50 PM \nVenue: AC04-LR-301 \nSpeaker: Laurence M. Ball \nTitle: "U.S. Inflation Since the Pandemic" \nAbstract: This work seeks to explain the behavior of US inflation since 2020. The starting point is a decomposition of headline inflation into core inflation\, as measured by the weighted median of industry inflation rates\, and deviations from core (“headline inflation shocks”). Over 2021-2022\, core inflation rose sharply as a result of three factors: a modest rise in long-term inflation expectations; a very tight labor market\, as indicated by a high ratio of job openings to unemployment; and pass-through from headline inflation shocks to core. The headline shocks\, which also contributed directly to headline inflation\, arose in industries such as energy and used cars. Since late 2022\, core and headline inflation have fallen most of the way back to their pre-pandemic levels because all three factors behind the rise in inflation have reversed. \nWe look forward to having you with us. \n– \nRegards \nDepartment of Economics \nAshoka University \nWeb: www.ashoka.edu.in \n 
URL:https://www.ashoka.edu.in/event/special-seminar-by-prof-laurence-m-ball/
END:VEVENT
END:VCALENDAR