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:20260402T024533Z
LOCATION:Level 2 Lobby
DTSTART;TZID=America/Los_Angeles:20250623T180000
DTEND;TZID=America/Los_Angeles:20250623T190000
UID:dac_DAC 2025_sess262_RESEARCH805@linklings.com
SUMMARY:Crosstalk-Aware Mapping for Optical Neural Networks
DESCRIPTION:Yuetong Fang (The Hong Kong University of Science and Technolo
 gy (Guangzhou)), Ziqing Wang (Northwestern University), and Renjing Xu (Th
 e Hong Kong University of Science and Technology (Guangzhou))\n\nBy levera
 ging wavelength division multiplexing (WDM) technology, optical neural net
 works (ONNs) built with micro-ring resonator (MRR) arrays have recently em
 erged as a powerful solution for accelerating the energy-intensive matrix-
 vector multiplication operations in artificial intelligence applications. 
 Despite their promise, the scalability of MRR-based computing systems is s
 everely limited by adjacent channel crosstalk. This challenge arises from 
 the inherent imperfections in MRR filtering, which lead to signal leakage 
 between neighboring channels. As the number of WDM channels increases, thi
 s accumulated adjacent channel interference intensifies, posing a substant
 ial obstacle to large-scale ONNs.  In this work, we propose the Crosstalk-
 Aware Mapping (CAM) strategy that mitigates this crosstalk by optimizing t
 he weight mapping scheme for MRR arrays. This is achieved through array-wi
 se weight reallocation and the optimization of phase-shift combinations on
  both row and column levels. Extensive experimental results on MRR-ONN sys
 tems substantiate the effectiveness of CAM, with average performance impro
 vements of 61.92%, 48.70%, and 63.08% on the MNIST, Fashion-MNIST, and CIF
 AR-10 datasets.\n\nTracks: DES5: Emerging Device and Interconnect Technolo
 gies\n\n
END:VEVENT
END:VCALENDAR