Frugal and Robust AI for Defence Advanced Intelligence

An important crosscutting need for Artificial Intelligence is to create technologies for trustworthy autonomous and frugal learning, i.e. the ability of a system to adapt and learn from its environment, including from user supervision, for a reasonable cost and without intervention from expert developers nor regression. Such technologies can be highly disruptive and have high impacts for many capabilities, especially when the information to manage is highly variable or unpredictable and high adaptability is needed. These technologies can also alleviate the current need to provide data to the system developers to get improvements depending on such data, which can be critical when the data is confidential, and is thus critical for defence. They can more generally enhance technological independence. Selected actions should include the organisation of technological challenges addressing well-defined goals in order to bootstrap and drive progress toward answering identified defence needs, while leveraging civil research and generating spill over effects. Within the FaRADAI project, current advances in AI technologies will be thoroughly researched in parallel with a detailed study of the main challenges imposed by a defence system. Aiming at significant breakthroughs in AI, the models will accelerate their wider application and deployment in defence systems increasing their impact and the overall performance.

M4D, acts as the project’s coordinator and scientific manager of FaRADAI. In addition, the group is leading the task related to the fusion of outcomes of different modalities in a complementary manner, towards the extraction of additional knowledge at next processing levels. Lastly, M4D is responsible to contribute to the design and development of methods that can be trained with the use of less or no annotated data via different learning paradigms.