BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-LIC-LOCATION:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260730T152640Z
LOCATION:Exhibit Hall
DTSTART;TZID=America/Los_Angeles:20260728T175400
DTEND;TZID=America/Los_Angeles:20260728T175500
UID:dac_DAC 2026_sess306_WIP3294@linklings.com
SUMMARY:Lighthouse RL: Sample-Efficient Circuit Optimization via Strategic
  Reset Points
DESCRIPTION:Mustafa Gürsoy, Stefan Uhlich, Ryoga Matsuo, Yağız Gençer, Aru
 n Venkitaraman, Chia-Yu Hsieh, Andrea Bonetti, and Lorenzo Servadei (Sony 
 AI)\n\nIn this paper, we introduce Lighthouse RL, a sample-efficient reinf
 orcement learning (RL) approach for analog circuit sizing. Traditional met
 hods lack generalization across different performance targets, while stand
 ard RL approaches waste resources exploring unpromising regions. Our metho
 d addresses these inefficiencies through a strategic reset strategy that i
 nitializes episodes from high-performing configurations discovered during 
 training, called "lighthouses". These states, which are closer to the targ
 et objectives, guide exploration toward promising regions. When compared t
 o RL and Bayesian optimization methods from the literature, we demonstrate
  the effectiveness of our approach on a 2D benchmark problem and on two an
 alog circuits, showing significant improvements in sample efficiency (up t
 o 1.72× faster), optimization performance (100% vs. 0-87% success rate), g
 eneralization (75% vs. 0-50% extrapolation success), and objective maximiz
 ation. This efficiency is particularly valuable for computationally expens
 ive black-box optimization problems, and our reset strategy can be used as
  a plug-and-play enhancement for any RL-based optimization approach.\n\nTr
 ack: Student\n\n
END:VEVENT
END:VCALENDAR