The role of Anomaly Detection in X-ray security imaging, as a supplement to targeted threat detection, is described; and a taxonomy of anomalies types in this domain is presented. Algorithms are described for detecting appearance anomalies, of shape, texture and density; and semantic anomalies of object category presence. The anomalies are detected on the basis of representations extracted from a convolutional neural network pre-trained to identify object categories in photographs: from the final pooling layer for appearance anomalies, and from the logit layer for semantic anomalies. The distribution of representations in normal data are modelled using high-dimensional, fullcovariance, Gaussians; and anomalies are scored according to their likelihood relative to those models. The algorithms are tested on X-ray parcel images using stream-of-commerce data as the normal class, and parcels with firearms present as examples of anomalies to be detected. Despite the representations being learnt for photographic images, and the varied contents of stream-ofcommerce parcels; the system, trained on stream-of-commerce images only, is able to detect 90% of firearms as anomalies, while raising false alarms on 18% of stream-of-commerce.
A review was conducted to identify possible applications of artificial intelligence and related technologies in the perpetration of crime. The collected examples were used to devise an approximate taxonomy of criminal applications for the purpose of assessing their relative threat levels. The exercise culminated in a 2-day workshop on 'AI & Future Crime' with representatives from academia, police, defence, government and the private sector. The workshop remit was (i) to catalogue potential criminal and terror threats arising from increasing adoption and power of artificial intelligence, and (ii) to rank these threats in terms of expected victim harm, criminal profit, criminal achievability and difficulty of defeat. Eighteen categories of threat were identified and rated. Five of the six highest-rated had a broad societal impact, such as those involving AI-generated fake content, or could operate at scale through use of AI automation; the sixth was abuse of driverless vehicle technology for terrorist attack.
Existing approaches to automated security image analysis focus on the detection of particular classes of threat. However, this mode of inspection is ineffectual when dealing with mature classes of threat, for which adversaries have refined effective concealment techniques. Furthermore, these methods may be unable to detect potential threats that have never been seen before. Therefore, in this paper, we investigate an anomaly detection framework, at X-ray image patch-level, based on: (i) image representations, and (ii) the detection of anomalies relative to those representations. We present encouraging preliminary results, using representations learnt using convolutional neural networks, as well as several contributions to a general-purpose anomaly detection algorithm based on decision-tree learning.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.