Search

On the occasion of the tenth anniversary of the RDA, and as we approach the end of our Belmont-funded project, PARSEC (www.parsecproject.org), we felt it was time to discuss the practicalities of collaboration, data and code sharing for publication and re-use across the globe. The PARSEC team–from five geographically-dispersed countries–has collaborated for four years on the collation and harmonisation of data and the development of new methods for sharing data and code as we investigate the socio-economic effects of nature conservation initiatives. We have had very profitable partnerships in this endeavour with several leading data infrastructure and research tool providers, including ORCID, DataCite, the RDA itself, and the World Data System. In this session representatives of the key data science infrastructures (ORCID, Scholix, Crossref, DataCite, the WDS, the Environmental Data Initiative) and users (representing the voice of marine conservation, machine learning, data for artificial intelligence, health and life sciences, social inequalities in health, the geoscience community and domain variations in open science and open data) discuss the challenges they face and their vision of the optimum path to the future., This project was conducted as part of the Belmont Forum PARSEC project, funded under the Collaborative Research Action (CRA) on Science-Driven e-Infrastructures Innovation (SEI), with funding from FAPESP, the ANR, JST and the NSF, with collaborators from Australia, and support from the synthesis centre CESAB of the French Foundation for Research on Biodiversity.

This talk was part of theNovember Narratives Twin-Talks - Reproducibility of ML Experiments and National Research Graph on 17 Nov 2022. It provides some insights from the paper ‘Mitigation Strategies to Improve Reproducibility of Poverty Estimations from Remote Sensing Images Using Deep Learning’ Machicao, J., Ben Abbes, A., Meneguzzi, L., Corrêa, P.L.P., Specht, A., David, R., Subsol, G., Vellenich, D., Devillers, R., Stall, S., Mouquet, N., Chaumont, M., Berti‐Equille, L., Mouillot, D., 2022. Mitigation Strategies to Improve Reproducibility of Poverty Estimations From Remote Sensing Images Using Deep Learning. Earth and Space Science 9.https://doi.org/10.1029/2022EA002379 You can watch the talk on: https://www.youtube.com/watch?v=gAmL40vtcTg This research is product of the PARSEC group funded by the Belmont Forum as part of its Collaborative Research Action (CRA) on Science-Driven e-Infrastructures Innovation (SEI) with support from FAPESP, JST, ANR and NSF, and thesynthesis centre CESAB of the French Foundation for Research on Biodiversity.&nbsp

In recent years efforts to develop good research practices have accelerated, and issues of Reproducibility and Replicability (R & R) in computer science experiments have gained attention. Due to the complexity of the approaches used, the reproducibility of deep learning (DL) experiments is a particular challenge. Without the use of DL technology, however, it would be impossible to solve problems which require the acquisition and analysis of enormous amounts of data across space and time. Future users could greatly benefit from the experience of creating and running such experiments, and the challenges to reproducibility of DL experiments should be overcome. To investigate the nature of DL reproducibility challenges in detail and to find solutions that might improve their reproducibility, we assessed three DL studies which analysed visual elements from satellite and street imagery to estimate poverty levels in human settlements. We systematically detailed difficulties with the datasets, methods, and workflows for each experiment and identified actions to militate common problems. Our target audience includes researchers who want to replicate an experiment, authors who want to report on an experiment, and reviewers who want to evaluate DL experiments. As a result of this assessment, we have developed two approaches to improve the reproducibility and replicability of DL experiments: (i) a quality checklist incorporating the relevant FAIR principles inter alia, and (ii) guidelines to reduce the likelihood of wasted effort when deciding whether an experiment is able to be reproduced in a feasible manner.

PARSEC Scientific results: Methodology and Case Studies. This session at the University of Sao Paulo's 6th Data Science Symposium was a presentation of PARSEC results to date. The organisers are grateful for funding from IUGS and UNESCO to support the attendance of early career researchers. • PARSEC execution. Introduction to the application of the PARSEC synthesis model with case studies from East Africa. Ali Ben Abbes (Foundation for Research on Biodiversity-CESAB) Research foci • Estimation of wealth index in Brazil: Pedro Luiz Pizzigatti Corrêa (Escola Politécnica da USP), Jeaneth Machicao and team • Socio-economic assessment and PAs in Australia: Alison Specht (University of Queensland, TERN), Germàn Gonzalez (AURIN), Michael Rigby (AURIN) • Mapping the socio-economic consequences of PAs in west coast USA: Jamie Trammell (Southern Oregon University) and Jeffrey Evans (The Nature Conservancy) • Pandemic effects on tourism: Yasuhisa Kondo (Research Institute for Humanity and Nature, RIHN)&nbsp

Undegraduate poster related to PARSEC project presented at the 28º SIICUSP. Use of scientific data of protected areas for analysis of its socioeconomic impact., Acknowledgments: The PARSEC project is funded by the Belmont Forum, Collaborative Research Action on Science-Driven e-Infrastructures Innovation.