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:3006\, Level 3
DTSTART;TZID=America/Los_Angeles:20250624T113000
DTEND;TZID=America/Los_Angeles:20250624T114500
UID:dac_DAC 2025_sess148_RESEARCH650@linklings.com
SUMMARY:Ensembler: Protect Collaborative Inference Privacy from Model Inve
 rsion Attack via Selective Ensemble
DESCRIPTION:Dancheng Liu (State University of New York) and Chenhui Xu, Ji
 ajie Li, Amir Nassereldine, and Jinjun Xiong (University at Buffalo)\n\nDu
 ring collaborative inference with a cloud, it is sometimes essential for t
 he client to shield its sensitive information. In this paper, we introduce
  Ensembler, an extensible framework designed to substantially increase the
  difficulty of conducting model inversion attacks for adversarial parties.
  Ensembler leverages selective model ensemble on the adversarial server to
  obfuscate its reconstruction. Our experiments demonstrate that Ensembler 
 can effectively shield images from reconstruction attacks when the client 
 keeps even just one layer, significantly outperforming baseline methods by
  up to 43.5% in structural similarity. At the same time, Ensembler only in
 curs 4.8% overhead during inference time.\n\nTopics: Security\n\nTracks: S
 EC1: AI/ML Security/Privacy\n\nSession Chairs: Adnan Siraj Rakin (Binghamt
 on University) and Ayesha Siddique (University of Maine)\n\n
END:VEVENT
END:VCALENDAR