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:20260728T173300
DTEND;TZID=America/Los_Angeles:20260728T173300
UID:dac_DAC 2026_sess317_RESEARCH2030@linklings.com
SUMMARY:On-Demand Multi-Task Sparsity for Efficient Large-Model Deployment
  on Edge Devices
DESCRIPTION:Lianming Huang and Haibo Hu (City University of Hong Kong), Qi
 ao Li (Mohamed bin Zayed University of Artificial Intelligence), Nan Guan 
 (City University of Hong Kong), and Chun Jason Xue (MBZUAI)\n\nSparsity is
  essential for deploying large models on resource constrained edge platfor
 ms. However, optimizing sparsity patterns for individual tasks in isolatio
 n ignores the significant I/O overhead incurred during frequent task switc
 hing. We introduce an on-demand multi-task sparsity framework specifically
  designed to minimize switching costs by maximizing parameter reuse. Unlik
 e monolithic approaches, we decompose weights into reusable block-granular
  units and align sparse structures across tasks to maximize overlap. By dy
 namically loading only the small differential set of blocks required for t
 he next task, our method effectively mitigates the cold-start latency inhe
 rent in traditional monolithic approaches. Experiments on a real-world aut
 onomous driving platform demonstrate that our framework achieves superior 
 switching efficiency, accelerating task switching by over 6.6× on average 
 compared to existing sparsity methods.\n\nTrack: EDA7-II. Physical Design 
 and Verification, Design\n\nSession Chairs: Christian Pilato (Politecnico 
 di Milano); Mohsen Imani (University of California, Irvine); Flavio Ponzin
 a (San Diego State University); Amin Farshidi (Cadence Design Systems, Inc
 .); Yonggan Fu (Georgia Institute of Technology, NVIDIA); Dimitrios Tsaras
  (Huawei, Hong Kong University of Science and Technology (HKUST)); Anh Ngu
 yen (Fujitsu Limited); Ruiyang Qin (Villanova University); Jiliang Zhang (
 Hunan University); Meng Li (Peking University); and Rassul Bairamkulov (Ad
 vanced Micro Devices (AMD))\n\n
END:VEVENT
END:VCALENDAR