BEGIN:VCALENDAR VERSION:2.0 X-WR-CALNAME:EventsCalendar PRODID:-//hacksw/handcal//NONSGML v1.0//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/New_York LAST-MODIFIED:20240422T053451Z TZURL:https://www.tzurl.org/zoneinfo-outlook/America/New_York X-LIC-LOCATION:America/New_York BEGIN:DAYLIGHT TZNAME:EDT TZOFFSETFROM:-0500 TZOFFSETTO:-0400 DTSTART:19700308T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU END:DAYLIGHT BEGIN:STANDARD TZNAME:EST TZOFFSETFROM:-0400 TZOFFSETTO:-0500 DTSTART:19701101T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT CATEGORIES:College of Engineering,Graduate Studies,Lectures and Seminars,Th esis/Dissertations DESCRIPTION:Faculty Supervisor:Dr. Gokhan Kul, Computer & Information Scien ce/Data Science Committee Members:Dr. Gavin Fay, SMAST Fisheries & Oceanog raphyDr. Firas Khatib, Computer & Information Science/Data ScienceDr. Asho kkumar Patel, Computer & Information Science/Data Science Abstract: Marine ecosystem models such as Atlantis valuable insight into multispecies inte ractions, spatial dynamics and fisheries management, but their high comput ational cost limits rapid scenario analysis and real-time decision making. This study presented a data-driven ecosystem emulator for the Northeast U .S. Atlantis model using automated machine learning approaches. A large -s cale dataset (~3.4 million records) spanning 1964-2020 was constructed, i ntegrating biomass of species functional groups, spatial polygons, tempora l indices and environments variables such as temperature and salinity. The emulator framework employed automated machine learning techniques, includ ing Random Forest and Extra tress regression, with model selection and hyp erparameter optimization performance using AutoML strategies. In addition, AutoKeras was utilized to explore neural network architectures in an auto mated manner, enabling data-driven model allocation, lagged variables to c apture ecological inertia and time-aware transformations. Model performanc e was evaluated using out-of-sample temporal validation, recursive back te sting, and ecological plausibility assessments. Result demonstrated string predictive performance, with species -level R2 values frequently exceedin ag 0.90 and overall model accuracy approaching 94%. The emulator achieved high computational efficiency, with end-to-end prediction completed in und er few seconds, substantiality reducing runtime compared to traditional At lantis simulations. This work established a scalable and efficient AutoML- driven alternative to process-based ecosystem model, enabling rapid biomas s estimation and supporting data-driven fisheries management and ecosystem analysis.  For further information please contact Dr. Gokhan Kul at gkul @umassd.edu \nEvent page: /events/cms/emulating-mar ine-ecosystem-dynamics-a-machine-learning-approach-for-biomass-prediction- and-forecasting.php\nEvent link: https://teams.microsoft.com/meet/21013462 2120814?p=HF1gXRarGipViGBpda X-ALT-DESC;FMTTYPE=text/html:

ÌÇÐÄlogoÈë¿Ú

Faculty Supervisor:
Dr. Go khan Kul\, Computer & Information Science/Data Science

\n

Committee M embers:
Dr. Gavin Fay\, SMAST Fisheries & Oceanography
Dr. Firas Khatib\, Computer & Information Science/Data Science
Dr. Ashokkumar Patel\, Computer & Information Science/Data Science

\n

Abstract: Mari ne ecosystem models such as Atlantis valuable insight into multispecies in teractions\, spatial dynamics and fisheries management\, but their high co mputational cost limits rapid scenario analysis and real-time decision mak ing. This study presented a data-driven ecosystem emulator for the Northea st U.S. Atlantis model using automated machine learning approaches. A larg e -scale dataset (~3.4 million records) spanning 1964-2020 was constructed \, integrating biomass of species functional groups\, spatial polygons\, temporal indices and environments variables such as temperature and salini ty. The emulator framework employed automated machine learning techniques\ , including Random Forest and Extra tress regression\, with model selectio n and hyperparameter optimization performance using AutoML strategies. In addition\, AutoKeras was utilized to explore neural network architectures in an automated manner\, enabling data-driven model allocation\, lagged va riables to capture ecological inertia and time-aware transformations. Mode l performance was evaluated using out-of-sample temporal validation\, recu rsive back testing\, and ecological plausibility assessments. Result demon strated string predictive performance\, with species -level R2 values freq uently exceedinag 0.90 and overall model accuracy approaching 94%. The emu lator achieved high computational efficiency\, with end-to-end prediction completed in under few seconds\, substantiality reducing runtime compared to traditional Atlantis simulations. This work established a scalable and efficient AutoML-driven alternative to process-based ecosystem model\, ena bling rapid biomass estimation and supporting data-driven fisheries manage ment and ecosystem analysis. 

\n

For further information please cont act Dr. Gokhan Kul at gkul@umassd.edu 

Event page:
Event link: < a href="https://teams.microsoft.com/meet/210134622120814?p=HF1gXRarGipViGB pda">https://teams.microsoft.com/meet/210134622120814?p=HF1gXRarGipViGBpda

DTSTAMP:20260425T100511 DTSTART;TZID=America/New_York:20260511T160000 DTEND;TZID=America/New_York:20260511T170000 LOCATION:Online/Teams (https://teams.microsoft.com/meet/210134622120814?p=H F1gXRarGipViGBpda) SUMMARY;LANGUAGE=en-us:Emulating Marine Ecosystem Dynamics: A Machine Learn ing Approach For Biomass Prediction And Forecasting UID:d3bfb5a104895db40e5f1096a7d05fda@www.umassd.edu END:VEVENT END:VCALENDAR