The project  represents an international collaborative project involving industry and academic
institutions to build new methodologies to improve AI based military applications in the call requirements (robustness, frugality,
resilience) and other (explainability, human control).
The project will be developed around four main outputs:
• Simulation framework: High performance computing capabilities for generative models, reducing complexity or producing
synthetic data (sensors, signals, …)
• Identification and development of use cases for testing new frugal methods co-designed with end-users.
• Definition of a metrics specifically designed for use cases measuring frugality (small data or quick adaptation to
uncertainities), robustness, resilience, explainability and performance assuring reproducibility and accountability in the design and
implementation.
• Development of experiments built on a common protocol for traceability and accountability of data and results, ensuring
their reproducibility.
Upon this analytical framework the project will provide new or transferred AI methodologies and techniques generated different
approaches (ML/DL, Symbolic, Metaheuristics, …) and research fields in AI (Knowledge representation and reasoning, Support to
Decision Making, Metaheuristics…) or transferred from other sectors (Security…)
The expected results in the project are mainly based in developing a novel AI approach based on different research fields and
disciplines (non-exhaustive list):
o Information, Control and Complexity theories
o Simulation and generative models
o Machine Learning and its approaches (semi or unsupervised, zero or few-shot, active, transfer and reinforcement)
o Knowledge representation and reasoning (knowledge graph, goal reasoning…)
o Metaheuristics
o Hybrid Artificial Intelligence (Neuro-symbolic approach)
o Support to decision (Bayesian, Markov…)
o Cybersecurity (conventional and adversarial ML attacks)