A Self-Supervised Framework for Space Object Behaviour CharacterisationHere, we demonstrate how self-supervised learning can simultaneously enable anomaly detection, motion prediction, and synthetic data generation from rich representations learned in pre-training. Our work therefore supports space safety and sustainability through automated monitoring and simulation capabilities. Learn more | Improving Object Detection by Modifying Synthetic Data with Explainable AIOur proof-of-concept results pave the way for fine, XAI-controlled curation of synthetic datasets tailored to improve object detection performance, whilst simultaneously reducing the burden on human operators in designing and optimising these datasets. Learn more | An AI blue team playbookIn a fiercely competitive landscape, we are deploying AI systems faster than they can be security tested and defended. Our playbook contains the blue teaming historical context, process, lessons learned and hypothetical examples, serving as a starting point for embedding security at the heart of AI-enabled systems. Learn more |
An AI red team playbookComplementing An AI Blue Team Playbook, this paper contains the red teaming historical context, process, and lessons learned, serving as a starting point for proactively identifying weaknesses, enhancing the overall performance, security, and resilience of AI-enabled systems. Learn more | Radar Pulse Deinterleaving with Transformer Based Deep Metric LearningWhen receiving radar pulses it is common for a recorded pulse train to contain pulses from many different emitters. In this paper, we define the problem and present metrics that can be used to measure model performance. This model achieves strong results in comparison with other deep learning models with an adjusted mutual information score of 0.882. Learn more | 2DSig-Detect: a semi-supervised framework for anomaly detection on image data using 2D-signaturesThe rapid advancement of machine learning technologies raises questions about the security of machine learning models. Using 2DSig-Detect for anomaly detection, we show both superior performance and a reduction in the computation time to detect the presence of adversarial perturbations in images. Learn more |
Artificial Intelligence in WargamingThis report explores two possible investment pathways for AI in wargaming: 1) narrow, specialised AI applications for the near-term, and 2) high-risk, high-reward AI investments. We conclude that the benefits AI can bring to wargaming could be significant, but there would be benefit in first introducing automation in specifically tactical or abductive wargames in the near term to manage risks. Learn more | SHARDeg: A Benchmark for Skeletal Human Action Recognition in Degraded ScenariosComputer vision (CV) models for detection, prediction or classification tasks operate on video data-streams that are often degraded in the real world, due to deployment in real-time or on resource-constrained hardware. Here we address this issue for SHAR by providing an important first data degradation benchmark on the most detailed and largest 3D open dataset, NTU-RGB+D-120, and assess the robustness of five leading SHAR models to three forms of degradation that represent real-world issues. Learn more | On-board Mission Replanning for Adaptive Cooperative Multi-Robot SystemsCooperative autonomous robotic systems have significant potential for executing complex multi-task missions across space, air, ground and maritime domains. But they commonly operate in remote, dynamic and hazardous environments. Fast, on-board replanning algorithms are therefore needed to enhance resilience. Here we define the Cooperative Mission Replanning Problem as a novel variant of multiple TSP with adaptations to overcome these issues, and develop a new encoder/decoder-based model using Graph Attention Networks and Attention Models to solve it effectively and efficiently. This work paves the way for increased resilience in autonomous multi-agent systems. Learn more |
