Recent developments in artificial intelligence (AI) have highlighted the significant potential of the technology to increase Defence capability including improving performance, removing humans from high-threat environments, reducing capability costs and achieving asymmetric advantage. However, significant work is required to ensure that introducing the technology does not result in adverse outcomes

Anomaly detection is the task of sorting data points into normal and anomalous classes. In a semi-supervised setting, there is only access to normal class samples during the training phase. Through generating and ranking an appropriate scalar quantity, these training samples can be used to calibrate the anomaly detector so that it produces a specified false alarm rate.

The Defence Science and Technology (DST) Land Capability Analysis (LCA) Future Technology Concept Exploration (FTCE) programme focusses on designing novel ways of operating to exploit and counter emerging technologies, and assessing the potential operational effectiveness of the conceptual and structural transformations.

In this report, a power series formulation of feature importance that explicitly identifies individual and interaction-type contributions is discussed.

This report thus investigates the subject of factor screening, for stochastic simulation models, and overviews several solutions to this problem. In particular, sequential bifurcation will be shown to provide a very efficient approach to the problem of factor screening, in comparison to standard one factor at a time classication methods.

In this report, we review the foundations of cross-validation and draw attention to common, but underappreciated, assumptions.

This paper describes a new fully automatic data driven real-time model of the morphology of Es, and describes the associated expected propagation characteristic that should reduce the need for manual intervention. It has been adapted to work with JORN ionospheric sounder data in real time and tested with years of data.

In this report we seek to evaluate the utility of dirichlet multinomial mixture (DMM) modelling of network metadata in roles such as source characterisation, detection of cyber security events, or volume filtering in support of same.