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:20240626T180033Z
LOCATION:3002\, 3rd Floor
DTSTART;TZID=America/Los_Angeles:20240627T111500
DTEND;TZID=America/Los_Angeles:20240627T113000
UID:dac_DAC 2024_sess160_RESEARCH1872@linklings.com
SUMMARY:Learn-by-Compare: Analog Performance Prediction using Contrastive 
 Regression with Design Knowledge
DESCRIPTION:Research Manuscript\n\nZihu Wang, Karthik Somayaji N.S., and P
 eng Li (University of California, Santa Barbara)\n\nThis paper introduces 
 Learn-by-Compare (LbC), a novel approach for analog performance modeling b
 y employing semi-supervised contrastive regression. LbC employs a deep neu
 ral network encoder to come up with latent representations of sizing solut
 ions by comparing similarity/dissimilarity of the underlying performance. 
 Leveraging two levels of transistor-level sizing data augmentation (DA), n
 amely LS-DA and GS-DA, LbC produces new data samples by employing design k
 nowledge. Experimental results highlight LbC's superior predictive accurac
 y compared to traditional regression methods. Offering a streamlined semi-
 supervised learning methodology, LbC effectively incorporates simple desig
 n knowledge and representation learning for efficient analog performance m
 odeling.\n\nTopic: EDA\n\nKeyword: Analog CAD, Simulation, Verification an
 d Test\n\nSession Chair: Markus Olbrich (Leibniz University Hannover)
END:VEVENT
END:VCALENDAR