BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/New_York BEGIN:DAYLIGHT DTSTART:20230312T030000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:EDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20221106T010000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:EST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20230503T105809Z UID:0307C635-0395-46B2-B2C8-4CBA71DF8862 DTSTART;TZID=America/New_York:20221114T110000 DTEND;TZID=America/New_York:20221114T120000 DESCRIPTION:Abstract: Underwater sound propagation is primarily driven by a non-linear forward model relating variability of the ocean sound speed pr ofile (SSP) to the acoustic observations (e.g. eigenray arrival times). Oc ean acoustic tomography (OAT) methods aim at reconstructing SSPs variation s (with respect to a reference environment) from changes of the acoustic m easurements between multiple source-receiver pairs. This presentation will review the performance of three different OAT methods: 1) model-based met hods (i.e. classical ray-based OAT using a linearized forward model)\, 2) data-driven methods (such as deep learning) to directly learn the inverse model and 3) a hybrid solution (i.e. the Neural Adjoint-NA-method) which c ombines deep learning of the forward model followed by a standard recursiv e optimization to estimate SSPs. Additionally\, synthetic SSPs were genera ted to augment the variability of the training set. These methods were tes ted with modeled ray arrivals calculated for a downward refracting environ ment with mild fluctuations of the thermocline. Idealized towed and fixed source configurations are considered. Results indicate that merging data-d riven and model-based methods can benefit OAT predictions depending on the selected sensing configurations and actual ray coverage of the water colu mn. But ultimately\, the robustness of OAT predictions depends on the dyna mics of the SSP variations.\n\nBiography: Dr. Sabra began at Tech in 2007 as an Assistant Professor. He emphasizes an interdisciplinary approach to research problems in different areas (acoustics\, structural health monito ring\, biomechanics and seismology) based on common wave propagation physi cs features. His research combines both theoretical and experimental studi es for the exploration of these research problems.\n\nFood: Pizza lunch to be provided.\n\nCo-sponsored by: Kevin Berman\; Lezheng Fang\n\nSpeaker(s ): Dr. Karim Sabra\, \n\nRoom: 3403\, Bldg: MRDC\, 771 Ferst Dr NW\, Atlan ta\, Georgia\, United States\, 30332\, Virtual: https://events.vtools.ieee .org/m/332144 LOCATION:Room: 3403\, Bldg: MRDC\, 771 Ferst Dr NW\, Atlanta\, Georgia\, Un ited States\, 30332\, Virtual: https://events.vtools.ieee.org/m/332144 ORGANIZER:kevin.berman@gtri.gatech.edu SEQUENCE:5 SUMMARY:IEEE ASA/SPS Lecture - Machine Learning Approaches for Ocean Acoust ic Tomography URL;VALUE=URI:https://events.vtools.ieee.org/m/332144 X-ALT-DESC:Description: <br /><p><strong>Abstract</strong>: Underwater soun d propagation is primarily driven by a non-linear forward model relating v ariability of the ocean sound speed profile (SSP) to the acoustic observat ions (e.g. eigenray arrival times). Ocean acoustic tomography (OAT) method s aim at reconstructing SSPs variations (with respect to a reference envir onment) from changes of the acoustic measurements between multiple source- receiver pairs. This presentation will review the performance of three dif ferent OAT methods: 1) model-based methods (i.e. classical ray-based OAT u sing a linearized forward model)\, 2) data-driven methods (such as deep le arning) to directly learn the inverse model and 3) a hybrid solution (i.e. the Neural Adjoint-NA-method) which combines deep learning of the forward model followed by a standard recursive optimization to estimate SSPs. Add itionally\, synthetic SSPs were generated to augment the variability of th e training set. These methods were tested with modeled ray arrivals calcul ated for a downward refracting environment with mild fluctuations of the t hermocline. Idealized towed and fixed source configurations are considered . Results indicate that merging data-driven and model-based methods can be nefit OAT predictions depending on the selected sensing configurations and actual ray coverage of the water column. But ultimately\, the robustness of OAT predictions depends on the dynamics of the SSP variations.</p>\n<p> &nbsp\;</p>\n<p><strong>Biography</strong>: Dr. Sabra began at Tech in 200 7 as an Assistant Professor. He emphasizes an interdisciplinary approach t o research problems in different areas (acoustics\, structural health moni toring\, biomechanics and seismology) based on common wave propagation phy sics features. His research combines both theoretical and experimental stu dies for the exploration of these research problems.<br /><br /><strong>Fo od:&nbsp\;</strong>Pizza lunch to be provided.</p> END:VEVENT END:VCALENDAR