Palestra IPMA DivRP - terça feira, 3/10/2023 11:30h - formato híbrido
Marta Rufino
Vimos por este meio anunciar ou relembrar, mais uma palestra no nosso ciclo de palestras do IPMA-DivRP.
Para os que planeiam assistir presencialmente, convidamos a estarem presentes na sala castanha do IPMA (sala Luis Saldanha).
Link para o evento, titulo, resumo e biografia em baixo (a completar).
Para os participantes por zoom, é fundamental colocarem o vosso nome e instituição quando entrarem no zoom.
Até lá !
Marta e Alexandra
__________
Dear colleagues,
We hereby announce or recall, another lecture in our cycle of IPMA-DivRP lectures.
People attending in person are invited to the brown IPMA room (Luis Saldanha room).
Link to event, title, abstract and biography below (to be completed).
For zoom participants, it is essential to put your name and institution when you enter the zoom.
See you then !
SAVE THE DATE
Lugar: Sala castanha, IPMA Algés
03/10/2023 11:30h PT (12:30h CET) **** TERÇA FEIRA****
Nome e afiliação: João Samarão (IPMA, Nova School of Science and Technology )
Title: Improving machine learning predictions when estimating fishing effort using high resolution spatio-temporal data
Abstract: Global fish production is forecasted to reach 200 Mt by 2029, putting enormous pressure on the ecosystem sustainability and marine stocks. In this study, high-resolution spatial-temporal data was used to attest how machine learning algorithms can be used to detect fishing activity. Several pre-processing and post-processing steps were applied to the data to improve the machine-learning predictions. Results illustrated that Random Forest, and Extreme Gradient Boosting would be the best algorithm to detect fishing activity with performances close to 99%. Furthermore, a brief explanation of future work using mostly spatial information is presented.
Biography: João Samarão is a master’s student in Analysis and Engineering of Big Data at Nova School of Science and Technology, in Almada. Currently developing his thesis at the Portuguese Institute of the Sea and Atmosphere (IPMA) supervised by Marta Rufino, focusing on a pipeline that has the objective of identifying fishing gear and detecting fishing activity by applying Neural Networks to spatial data. Since 2022 he has been working for IPMA with spatio-temporal high-resolution data to implement machine learning methodologies to improve Small-Scale Fisheries management and estimate fishing effort.
_____________________________________
Portuguese Institute for the Sea and the Atmosphere (IPMA)
Centre of Statistics and its Applications (CEAUL)
Faculty of Sciences, Univ. of Lisbon, Portugal