BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260202T201803Z
LOCATION:261-262-265-266
DTSTART;TZID=America/Chicago:20251120T135200
DTEND;TZID=America/Chicago:20251120T141500
UID:submissions.supercomputing.org_SC25_sess162_pap517@linklings.com
SUMMARY:Generative Latent Diffusion for Efficient Spatiotemporal Data Redu
 ction
DESCRIPTION:Xiao Li, Liangji Zhu, Anand Rangarajan, and Sanjay Ranka (Univ
 ersity of Florida)\n\nGenerative models have demonstrated strong performan
 ce in conditional settings and can be viewed as a form of data compression
 , where the condition serves as a compact representation. However, their l
 imited controllability and reconstruction accuracy restrict their practica
 l application to data compression. In this work, we propose an efficient l
 atent diffusion framework that bridges this gap by combining a variational
  autoencoder with a conditional diffusion model. Our method compresses a s
 mall number of keyframes into latent space and uses them as conditioning i
 nputs to reconstruct the remaining frames via generative interpolation, el
 iminating the need to store latent representations for every frame. This a
 pproach enables accurate spatiotemporal reconstruction while significantly
  reducing storage costs. Experimental results across multiple datasets sho
 w that our method achieves up to 10× higher compression ratios than rule-b
 ased state-of-the-art compressors such as SZ3, and up to 63% better perfor
 mance than leading learning-based methods under the same reconstruction er
 ror.\n\nTag: Algorithms, Applications, State of the Practice\n\nRecording:
  Livestreamed, Recorded\n\nRegistration Category: Technical Program Reg Pa
 ss\n\nSession Chair: Shaikh Arifuzzaman (University of Nevada, Las Vegas)\
 n\n
END:VEVENT
END:VCALENDAR