- optimizing AI methods for research in biodiversity,
- designing innovative prediction models and indicators,
- designing hybrid AI techniques.
The challenge is open to scientific communities in the fields of AI and biodiversity in France and in AFD’s partner countries in Africa.
AI research to better predict changes in biodiversity
Furthering knowledge about biodiversity and its changes is a major scientific and societal challenge that is at the heart of the National Biodiversity Plan. Faced with the complexity of the interactions between ecosystems and the interactions and feedback they involve, and given the heterogeneity of the data, current indicators are limited in their ability to predict changes in biodiversity. AI can offer a significant contribution through new, dedicated solutions. France has implemented a national AI research program to address this major challenge.
Data structuring, a crucial issue
The difficulty of structuring data has been identified as one of the primary hinderances to understanding biodiversity by the “biodiversity IPCC”, the IPBES (Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services). The IA-Biodiv Challenge has therefore been designed to mobilize the expertise of scientific communities in AI and biodiversity to build a dedicated virtual research environment.
“IA-Biodiv-Net”: to provide datasets and AI modules for research
After the evaluation and selection of the proposals by an international committee, the chosen research teams will work collaboratively over four years on the design and development of this environment.
Called “IA-Biodiv-Net” and developed by and for researchers, it will host datasets and AI modules developed by the consortia taking part in the Challenge, as well as evaluation modules set up by the operational consortium. It will comply to the extent possible with the FAIR principles (Findability, Accessibility, Interoperability, Reusability) and will be designed to be added to outside of the challenge and accessible to all interested scientific communities. It will thus contribute in the long term to the distribution of sustainable datasets for future scientific work in AI and the field of biodiversity.
Collective reflection on scientific challenges and research approaches
The Challenge has been set up in the framework of a specific ANR funding instrument that allows different scientific or technological approaches to a single challenge to be explored at the same time. It aims to compare and share the ideas, approaches and work of the funded consortia and promote the creation of common benchmarks between scientific communities from different backgrounds.
In the IA-Biodiv Challenge, the scientific monitoring of the funded consortia will be coordinated by an operational consortium and their work will be evaluated during annual meetings aiming to characterize and share the IA solutions being studied. This process will also facilitate cooperation among the different consortia.
A Challenge backed by €4.8 million to support research
A total of €4.8 million will be put into funding this scientific challenge, through a partnership between the ANR (€1.5M) and AFD (€1.5M) and a donation by Facebook Artificial Intelligence Research (€1.8M).
AFD receives the technical support of Expertise France to implement this Challenge. Expertise France is a key actor in international technical cooperation, through which it is possible to finance and support African consortia.