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:20260402T024534Z
LOCATION:3003\, Level 3
DTSTART;TZID=America/Los_Angeles:20250624T161500
DTEND;TZID=America/Los_Angeles:20250624T163000
UID:dac_DAC 2025_sess123_RESEARCH1194@linklings.com
SUMMARY:EVA: An Efficient and Versatile Generative Engine for Targeted Dis
 covery of Novel Analog Circuits
DESCRIPTION:Jian Gao (Northeastern University), Weimin Fu and Xiaolong Guo
  (Kansas State University), Weidong Cao (George Washington University), an
 d Xuan Zhang (Northeastern University)\n\nAnalog circuit design has tradit
 ionally depended on manual expertise, slowing the discovery of novel topol
 ogies essential for advanced technologies like AI, 5G/6G, and quantum comp
 uting. While AI-driven methods have accelerated hardware design workflows,
  most of them focus on topology synthesis, often reusing known structures 
 to achieve specific goals. The challenge of discovering entirely new, high
 -performance topologies remains largely underexplored due to its abstract 
 nature. In this work, we introduce EVA, an efficient and versatile generat
 ive engine for discovering novel analog circuit topologies. EVA employs a 
 bottom-up generation framework, using a decoder-only transformer to sequen
 tially predict device pin connections and create diverse circuits from scr
 atch. Pretraining on unlabeled circuit topologies builds foundational know
 ledge about circuit connectivity, achieving baseline discovery efficiency 
 by generating valid circuits and reducing performance-labeled samples need
 ed in fine-tuning. For targeted discovery of high-performance designs, EVA
  leverages two fine-tuning strategies—proximal policy optimization (PPO) a
 nd direct preference optimization (DPO)—to further enhance discovery effic
 iency for relevant, high-performing topologies. Experimental results acros
 s various circuit types highlight EVA's strengths in validity, novelty, ve
 rsatility, and both training sample and discovery efficiency.\n\nTopics: D
 esign\n\nTracks: DES4: Digital and Analog Circuits\n\nSession Chairs: Ioan
 nis Savidis (Drexel University) and Kishor Kunal (Nvidia)\n\n
END:VEVENT
END:VCALENDAR