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:
DTSTART;TZID=America/Los_Angeles:20260728T175100
DTEND;TZID=America/Los_Angeles:20260728T175200
UID:dac_DAC 2026_sess317_RESEARCH1899@linklings.com
SUMMARY:ACE-NAS: A Zero-Cost NAS Framework for Co-Design of Architecture a
 nd Encoding in Spiking Neural Networks
DESCRIPTION:Mingyu Qin (Sun Yat-Sen University) and Liangming Fang, Wei Li
 u, Jinghai Wang, and shanlin xiao (Sun Yat-sen University)\n\nSpiking Neur
 al Networks (SNNs) promise exceptional energy efficiency for neuromorphic 
 computing through event-driven processing. However, unlocking their full p
 otential requires navigating a\ncomplex, strongly coupled design space of 
 network topology and\ntemporal neural encoding. Existing SNN Neural Archit
 ecture Search\n(NAS) frameworks typically decouple these dimensions, focus
 ing\nexclusively on topology search while relying on manual, fixed encodin
 g schemes. This limitation leaves a vast portion of the design\nspace unex
 plored, resulting in sub-optimal energy-accuracy trade-offs. To bridge thi
 s gap, we present ACE-NAS, the first zero-cost\nNAS framework that automat
 es the co-design of SNN architecture\nand encoding mechanisms. Addressing 
 the challenge of "encoding-blind" proxies, we introduce Jacob_cov, a novel
  Jacobian-based\nspectral metric that efficiently quantifies the temporal 
 discriminability of different encoding schemes without training. ACE-NAS\n
 integrates Jacob_cov with structural proxies (ZiCo) into a hardware-aware,
  multi-objective evolutionary search strategy. On CIFAR-10,\nACE-NAS achie
 ves 92.95% accuracy, effectively identifying Pareto-optimal designs that b
 alance high performance with minimal spike\nactivity. By automating the jo
 int optimization of structure and\ndynamics, ACE-NAS delivers an orders-of
 -magnitude reduction\ncompared to standard training-based NAS and a 7× spe
 edup over\nstate-of-the-art efficient SNN-NAS methods (e.g., AutoSNN).\n\n
 Track: EDA7-II. Physical Design and Verification, Design\n\nSession Chairs
 : Christian Pilato (Politecnico di Milano); Mohsen Imani (University of Ca
 lifornia, Irvine); Flavio Ponzina (San Diego State University); Amin Farsh
 idi (Cadence Design Systems, Inc.); Yonggan Fu (Georgia Institute of Techn
 ology, NVIDIA); Dimitrios Tsaras (Huawei, Hong Kong University of Science 
 and Technology (HKUST)); Anh Nguyen (Fujitsu Limited); Ruiyang Qin (Villan
 ova University); Jiliang Zhang (Hunan University); Meng Li (Peking Univers
 ity); and Rassul Bairamkulov (Advanced Micro Devices (AMD))\n\n
END:VEVENT
END:VCALENDAR