Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 3
Encontrada(s) 1 página(s)

DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Can we harmonize the monitoring of plants and pollinators?

  • Porcher, Emmanuelle
  • Bonnet, Pierre
  • Damgaard, Christian
  • De Frenne, Pieter
  • Deguines, Nicolas
  • Ehlers, Bodil K.
  • Frei, Jerome
  • García, María B.
  • Gros, Clément
  • Jandt, Ute
  • Joly, Alexis
  • Martin, Gabrielle
  • Michez, Denis
  • Pescott, Oliver L.
  • Roth, Tobias
  • Waller, Donald
Meeting report on the symposium ‘New solutions to monitor plants, pollinators and their interactions in a changing world’, held at Collège de France and Museum national d’Histoire naturelle, Paris, France, 23–24 May 2024., The authors thank Collège de France (Chaire ‘Biodiversité et Ecosystèmes’, Fondation Jean-François et Marie-Laure de Clermont-Tonnerre) and Muséum national d'Histoire naturelle for funding the symposium, as well as the thousands of citizen scientists who make plant and pollinator monitoring possible. PB and AJ also acknowledge HORIZON EUROPE European Research Council (GUARDEN-101060693, MAMBO-101060639), and Agence Nationale de la Recherche (Pl@ntAgroEco-ANR-22-PEAE-0009). OLP acknowledges Natural Environment Research Council award no. NE/R016429/1 as part of the UK Status, Change and Projections of the Environment (UK-SCAPE) program delivering National Capability and Joint Nature Conservation Committee funding to the NPMS. References, Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

From Presence-Only to Abundance Species Distribution Models Using Transfer Learning

  • Bourel, Benjamin
  • Joly, Alexis
  • Servajean, Maximilien
  • Bettinger, Simon
  • Sanabria-Fernández, José Antonio
  • Mouillot, David
Species Distribution Models based on Convolutional Neural Networks (CNN-SDMs) have recently emerged, demonstrating greater effectiveness than traditional SDMs in several contexts. A limited number of studies, however, have focused on species abundance patterns, as the datasets available for this purpose are generally too small to effectively learn a deep learning model with millions of parameters. Our study demonstrated that CNN-SDMs can circumvent the small sample size of species abundance datasets through the combined use of a large presence-only species dataset and transfer learning to significantly improve the performance of abundance-based CNN-SDMs. Applied to Mediterranean coastal fishes, our approach significantly improves the abundance prediction performance of CNN-SDMs, with average gains of 35% (D-squared regression score). This allows CNN-SDMs to perform better than classical SDMs in abundance prediction, with average gains of 10%. These gains are stemming from enhanced abundance predictions for rare species and where widespread species are locally rare., This work was supported by European Union's Horizon research and innovation programme—MAMBO project (101060639), European Union's Horizon research and innovation programme—GUARDEN project (101060693), IA-Biodiv ANR—FISH-PREDICT project (ANR-21-AAFI-0001-01), Biodiversa+—BioBoost+ project (EU grant agreement 101052342), Junta de Andalucía Postdoctoral Fellowship (DGP_POST_2024_00757)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Otros (Other)

Beniofh/CNN_SDM_and_RF_for_Fish_2024

  • Bourel, Benjamin
  • Joly, Alexis
  • Servajean, Maximilien
  • Bettinger, Simon
  • Sanabria-Fernández, José Antonio
  • Mouillot, David
The aim of this repository is to make available the scripts and data needed to reproduce the experiments in our article on species distribution models (SDM). More specifically, these experiments use SDMs based on Convolutional Neural Networks (CNN-SDMs) and SDMs based on Random Forest (RF)., The downloadable data consists of three .zip files :, The inputs.zip (approximately 7 Go) contains the basic input data required to run most of the scripts. These data are the environmental rasters for all fish counts, some environmental data for presence-only fish data, tables of fish species data (presence-only and counts), and the outputs of the training models used in the article (training metadata and trained model weights)., The outputs.zip (approximately 500 Mo) contains examples of script outputs from the GitHub repository. Some of these outputs are required to run the scripts in the scr_metrics_and_figures folder in the GitHub repository., The complementary_inputs.zip (approximately 10 Go) contains additional input data in order to run the scripts cnn_sdm_presence_only.py and pred_cnn_sdm_presence_only.py (see part III.2) that cannot be run on standard machines. These are scripts that require a lot of computing resources and which must run on supercomputers. These additional data are all environmental rasters associated with the presence of fish., This work has been mainly funded by the IA-Biodiv ANR project FISH-PREDICT (ANR-21-AAFI-0001-01). It was also partially funded by the European Union’s Horizon research and innovation program under grant agreement No 101060639 (MAMBO project) and No 101060693 (GUARDEN project). This work was granted access to the HPC resources of IDRIS under the allocation 2022- AD011013891 made by GENCI »., Peer reviewed