Learning from the PARSEC project: deep learning in a socio-economic-conservation case study. How can we best ensure our outputs are FAIR and rewarded?
Abstract
The UNESCO Open Science recommendations (2021) exhort us to make multilingual scientific knowledge openly available [findable], accessible and reusable, to increase scientific collaborations and sharing of information for the benefits of science and society, and to open the processes of scientific knowledge creation, evaluation and communication to societal actors beyond the traditional scientific community. This is all very laudable, but harder to achieve in practice and, within the researcher community, as getting credit for many of these actions is slow to arrive. In a project, PARSEC (www.parsecproject.org) funded by the Belmont Forum (https://www.belmontforum.org/) we have been linking research thinking and technological innovation to accelerate the discovery of data and open science, with an objective to create solutions from potentially disruptive innovations.
The main scientific plank of PARSEC has been to use machine and deep learning to detect the socio-economic effects of protected area creation in four countries, Brazil, France, Japan, the USA, and as an add-on, Australia and East Africa. The theoretical value of this work is to identify positive situations where conservation actions have enhanced well-being. We were inspired by some publications in Africa (Jean et al., 2016 and Yeh et al., 2020), that used time series of remote sensing and census data. Parallel with this is the achievement of open science, sharing our synthesised data and code (we are solely confining ourselves re-using data and building on existing code) for others to use following the four principles of FAIR, while ensuring the benefits of that work comes back the researchers.
We have discovered many stumbling blocks in the re-use of data and code, and in this talk I will cover some deep learning experiments that have informed us about what would help make FAIR achievable. In particular we have proposed some ways reproducibility (the ‘R’ in FAIR) can be improved, from the user perspective and from the producer’s. Allied with this is a consideration of ways the researcher (that is ‘you’!) can be rewarded for engaging in what is commonly thought of as an after-thought to the ‘main game’. In this way we really might be able to build on the shoulders of giants.
Biography
Alison Specht has focussed over the last 10 years on facilitating interdisciplinary groups to tackle complex environmental problems using existing data. Her major interests, apart from her domain activities as an environmental scientist, are to improve data management and preservation, data sharing and re-use. From 2009 to 2014 she established the first synthesis centre in the southern hemisphere, the Australian Centre for Ecological Analysis and Synthesis, a facility of the Terrestrial Ecosystem Research Network (www.tern.org.au). From 2015 to 2018 she was Director of FRB-CESAB, the CEntre for the Synthesis and Analysis of Biodiversity in France. Alison was a member of a DataONE (www.dataone.org) Working Group in the USA from 2010-2019. She is a member of several RDA interest groups. She is a co-leader of PARSEC (www.parsecproject.org), a multi-national project funded under the Belmont Forum CRA for Science-driven e-infrastructures innovation for the enhancement of transnational, interdisciplinary and transdisciplinary data use in environmental change research.
HDR Candidates: Please email the seminar coordinator (usually the person that introduces the guest speaker) that you are in attendance.