BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20250626T234540Z
LOCATION:B301
DTSTART;TZID=America/New_York:20241118T115000
DTEND;TZID=America/New_York:20241118T121000
UID:submissions.supercomputing.org_SC24_sess746_ws_esp103@linklings.com
SUMMARY:Optimizing Uncertainty Estimation on Scientific Visualizations Usi
 ng Learning Models
DESCRIPTION:Erik Pautsch (Loyola University, Chicago; Argonne National Lab
 oratory (ANL)); David Guerrero-Pantoja and Clara Almeida (University of Ca
 lifornia, Santa Cruz; Argonne National Laboratory (ANL)); Maria Pantoja (C
 alifornia Polytechnic State University, San Luis Obispo; Argonne National 
 Laboratory (ANL)); Silvio Rizzi (Argonne National Laboratory (ANL)); and G
 eorge Thiruvathukal (Loyola University, Chicago; Argonne National Laborato
 ry (ANL))\n\nScientific visualizations (SciVis) convert numerical and spat
 ial data into images, enabling deeper insights into complex phenomena. Rec
 ent advancements in machine learning, particularly Deep Learning (DL), hav
 e significantly enhanced SciVis. By combining classical numerical techniqu
 es with DL, we can achieve the accuracies needed for real-world applicatio
 ns. However, DL models often exhibit undue confidence, especially with out
 -of-distribution (OOD) inputs, leading to misclassification with high conf
 idence scores. To address this, we enhance DL models by quantifying uncert
 ainty and enabling selective classification, allowing models to abstain fr
 om predictions when uncertainty is high. This approach outputs a predictio
 n distribution, guiding users on when to seek human intervention. We evalu
 ate this method across three tasks: airfoil pressure and velocity predicti
 on using a Reynolds-Averaged Navier-Stokes (RANS) model, image classificat
 ion with the ImageNet1K dataset, and digit recognition using the MNIST dat
 aset.\n\nTag: Applications and Application Frameworks, Algorithms, Perform
 ance Evaluation and/or Optimization Tools\n\nRegistration Category: Worksh
 op Reg Pass\n\nSession Chairs: Tiernan Casey (Sandia National Laboratories
 ) and Antigoni Georgiadou (Oak Ridge National Laboratory (ORNL))\n\n
END:VEVENT
END:VCALENDAR