BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260625T133337Z
LOCATION:Bldg. 6 - 001 - Plenary Room
DTSTART;TZID=Europe/Stockholm:20260630T123400
DTEND;TZID=Europe/Stockholm:20260630T123500
UID:submissions.pasc-conference.org_PASC26_sess129_posC109@linklings.com
SUMMARY:ACMP06 - Improving Deep Learning Based Seismic Inversion with Onli
 ne Augmentation
DESCRIPTION:Lucas Souza (Federal University of São Carlos)\n\nThis study i
 nvestigates the application of data augmentation techniques to optimize De
 ep Learning-based Seismic Inversion (DLI), aiming to overcome the scarcity
  of labeled datasets in the industry. Using the OpenFWI benchmark, the stu
 dy evaluates four incremental strategies: horizontal flipping, synthetic d
 ata generation via diffusion models, an innovative online augmentation loo
 p, and a hybrid DLI-FWI approach. The online augmentation, which generates
  physically consistent input-output pairs in real-time through forward mod
 eling, proved more efficient and effective than diffusion-based methods, s
 ignificantly reducing the Mean Absolute Error (MAE). The final methodology
 , which integrates network predictions with traditional Full-Waveform Inve
 rsion (FWI) refinement for complex geological cases, secured 12th place gl
 obally in the 2025 Geophysical Waveform Inversion competition.\n\nSession 
 Chair: Tobias Hodel (University of Bern, Switzerland)\n\n
END:VEVENT
END:VCALENDAR