Projects
Projects: Custom Search |
||
| Reference Number | UKRI2696 | |
| Title | FRAME-FM | |
| Status | Started | |
| Energy Categories | Renewable Energy Sources 15%; Other Cross-Cutting Technologies or Research (Energy Models) 15%; Not Energy Related 70%; |
|
| Research Types | Basic and strategic applied research 100% | |
| Science and Technology Fields | PHYSICAL SCIENCES AND MATHEMATICS (Statistics and Operational Research) 30%; PHYSICAL SCIENCES AND MATHEMATICS (Computer Science and Informatics) 40%; ENVIRONMENTAL SCIENCES (Earth Systems and Environmental Sciences) 30%; |
|
| UKERC Cross Cutting Characterisation | Not Cross-cutting 50%; Systems Analysis related to energy R&D (Energy modelling) 50%; |
|
| Principal Investigator |
Alberto Arribas National Oceanography Centre |
|
| Award Type | Standard | |
| Funding Source | EPSRC | |
| Start Date | 07 October 2025 | |
| End Date | 07 April 2026 | |
| Duration | 6 months | |
| Total Grant Value | £512,369 | |
| Industrial Sectors | Unknown | |
| Region | South East | |
| Programme | Artificial Intelligence and Robotics | |
| Investigators | Principal Investigator | Alberto Arribas , National Oceanography Centre |
| Other Investigator | Lily Gouldsbrough , Centre for Ecology & Hydrology Andrew Kingdon , British Geological Survey |
|
| Web Site | ||
| Objectives | ||
| Abstract | FRAME-FM aims to enable the fast and easy processing of diverse, complex and very large (peta-byte scale) environmental datasets by end-users in economic sectors including energy, food, finance and logistics. We will achieve this aim by delivering a framework that facilitates the development of Foundation Models (FMs): machine learning (ML) models that can encapsulate the information contained in large datasets – such as those existing in the data archives of NERC’s Environmental Data Service (EDS) – and can be fine-tuned to perform specialised tasks(1). FMs can greatly simplify the process of developing solutions to real-world challenges because they can be tailored to address specific tasks and can be easier to interact with than processing large raw datasets. FMs can contribute to the democratisation of data access, particularly benefitting end-users who may lack the computing resources and specialised knowledge required to process peta-byte scale datasets (or develop FMs) in the first place. However, creating environmental FMs is currently difficult because producers of environmental data, such as NERC EDS centres, lack the required digital infrastructure. The framework developed by the FRAME-FM project will include the software infrastructure, AI tools and standardised workflows necessary for NERC EDS. The FRAME-FM framework will be open-source and, although developed for environmental datasets, it will be designed to be generalisable – a key design principle of FRAME-FM will be to enable all data producers (including NERC centres) to deploy the framework directly on top of existing data archives – therefore benefitting other areas of science. The FRAME-FM project will evaluate existing state-of-the-art approaches to develop FMs, conduct user research to co-design the framework (ensuring its suitability for the environmental research community), consider key aspects such as security, explainability and environmental sustainability, build the software components necessary to test the framework, and develop at least one proof-of-concept relevant to end-users (e.g. renewable energy generation, food security, early warning systems for property protection) using data already available in the EDS archives. FRAME-FM will deliver two key outputs: A software infrastructure framework that enables and facilitates the development of FMs from big environmental datasets At least one practical demonstration of the framework to train AI Models using approaches typically used in FM development In addition, FRAME-FM will directly skill up multi-disciplinary researchers across EDS centres, and help speed up the adoption and use of FMs across public and private sector in a fair and inclusive way, contributing to make the UK a world-leader in the use of AI for environment. (1) Bommasani, Rishi, et al. "On the opportunities and risks of foundation models." arXiv preprint arXiv:2108.07258(2021) | |
| Data | No related datasets |
|
| Projects | No related projects |
|
| Publications | No related publications |
|
| Added to Database | 07/01/26 | |
