International Journal of Intelligent Defence Support Systems (5 papers in press)
Weapons Effects Modelling for Safety Applications
by Graeme Anderson, David Reed, Simon Di Nucci, Anthony Kwong, Janet Wilson
Abstract: Weapon effects modelling has many applications, but in this paper we focus on how advanced computer models can be used to assess the effectiveness of blast shielding structures and materials, for example to improve the safety of weapons production. Safety laws are changing from prescriptive, solution-based regulations to a risk-management-based approach, so being able to estimate the performance of protective measures allows us to produce directly-relevant safety evidence. Such an approach can be used to show that risks are reduced So Far As Is Reasonably Practicable (SFAIRP) in order to demonstrate compliance with the WHS Act, 2011.
This paper first describes how a variety of computer-based models for the prediction of the development and propagation of blast pressures can be used to provide a detailed assessment of the risks associated with explosions in complex environments, with particular reference to injury assessment - something that is sometimes impractical with simpler techniques. We then show how some of these techniques have been used in two real-world examples to successfully support safety case development. Finally, we extrapolate from the examples, showing how these techniques could be applied to assess weapons safety in different environments, or for risk assessment in defensive or offensive scenarios.rn
Keywords: Blast; fragment; structural response; injury; safety applications; risk-based management; SFAIRP.
Face Recognition under Occlusion for User authentication and Invigilation in Remotely distribute on-line assessments
by Niloofar Tavakolian, Azadeh Nazemi, Iain Murray
Abstract: Generally, running assessment requires many resources. Remote assessments solve local evaluations issues and eliminate some limitations. However, running the remote assessment service strictly is security dependent. Multi-Factor Electronic Authentication (MFA) by inserting a protection layer on top of username and password addresses the monitoring and security issues of remote assessment. On the result, users require passing two security levels to access online assessment services. They confirm what they know using username/password and verify what they have using biometric factors. This research proposes a method to manage aspects of distance digital learning in the assessment area of the learning organization. This study has focused on the face as a second biometric factor. Since Face recognition systems have been affected by variations of conditions such as expression, pose, occlusion, and illumination. The most highlighted challenge of this research was finding a solution to address uncontrolled conditions in face recognition. Obtained results of this project indicate reasonable accuracy to address the issue of occlusion using AR, MUCT and UMB Datasets. This research tried to address the issue regarding occlusion in face recognition using deep learning and comparing to the previous approach based on feature extraction(shallow method). Each of which has its own restrictions and advantages. The shallow method implemented using two different features HOG and Gabor. However, the shallow method can improve slightly, still cannot reach the Structures Sparse Representation based Classification (SSRC) the best member of the SRC family as state-of-the-art. The shallow method accuracy improvement includes Histogram of Oriented Gradient(HoG) by 4%, in comparison to Gabor SRC(GSRC) method and by 9% using Gabor as opposed to GSRC. On the other hand, this shallow method faced of the restriction regarding single sample training dataset. Additionally, due to the nature of the online authentication, occlusion cannot be predictable but shallow method can handle occlusion issue in the lack of occlusion dictionaries and sufficient training sample. Since the shallow method can address the limitations of this research, then to implement deep method initially preprocessing based on shallow method is performed and then modified ResNet is used to be able to improve accuracy comparing the best member of the SRC family by 3% on average.
Keywords: Face Recognition,Machine Learning,Occlusion,Illumination
Multi-Factor Authentication,Remote Assessment,Identification,Verification,
Expression; Face Detection,Deep Learning.
Research Studies on Human Cognitive Ability
by Ganesh Ram Sinha, K. Srujan Raju, Raj Kumar Patra, Win Aye, Thuzar Khin
Abstract: Research on cognitive ability of human brain can be seen now in literature that substantiates research scope and future directions in the field of human brain and its cognitive ability. This paper studies on research articles and work that highlighted various studies; surveys; and comparison on human brain research and its impact on cognitive ability for various artificial intelligence (AI) based applications. The research studies result several important questions on machine learning and impact of cognitive ability.
Keywords: Cognitive ability (CA); human brain; machine learning (ML); artificial intelligence (AI); deep learning (DL).
Opportunities with visualization linked with virtualized data: A case study
by Jyoti Prakash Mishra, Mishra Sambit Kumar, Mishra Anil Kumar
Abstract: The concept of virtualization linked with the fast growing network associated with the physical objects and IP address for internet connectivity may be a big challenge in the present society. Being associated with the objects and other Internet-enabled devices it may refer to the numerous physical devices around the world connected with internet for collecting and sharing data. Specifically, the technology associated with Internet of Things (IoT) may be viewed as the network of physical objects embedded with electronics, software, sensors, and network connectivity enabling to share and the exchange data linked among the physical and cyber space. It may also allow objects to be sensed and controlled remotely across existing network infrastructure,. It may also create opportunities for direct integration between the physical world and computer-based systems, and resulting in improved efficiency, accuracy and economic benefit. It has also been analyzed that by the year 2020, there will be over 50 billion connected devices and 6.58 devices connected per person. As it may focus on basic concept associated with smart data, it may be a major constraint towards processing the virtualized data. Accordingly, it may also be thought of connected devices to transfer data among one another in order to optimize the performance automatically. In this regard, processing the virtualized data during the application as well as during communications may also be a major challenge. Therefore, it has been proposed to analyze the opportunities as well as performance associated with virtualized data to simplify and strengthen human activities as well as expertise.
Keywords: Big Data; IoT; Virtualization; Sensor; Actuator; Data Center; Cluster.
A decision support web service for allocating assets in counter-piracy operations given periodic environmental forecast updates
by Francois-Alex Bourque, Raffaele Grasso, Raul Vicen-Bueno, Giampaolo Cimino, Paolo Braca, John Osler
Abstract: The rise of piracy in the Gulf of Aden and off the coast of Somalia prompted several nations to direct their navies to conduct counter-piracy operations in that region of the world. A major challenge with such operations is the sheer size of the area to patrol. One way to address this problem is by noting that pirates are more likely to operate under favourable environmental conditions and to focus patrols in areas where such conditions prevail by generating maps of piracy risk based on environmental forecast. Building on this notion, this contribution presents a system that was developed and implemented to help the decision makers plan operations over the course of days, weeks or months. Planning over an arbitrary time horizon is achieved by automatically re-allocating available assets with each new environmental forecast update, overcoming the inherent limitation of a single forecast (usually extending out to a few days) and resulting in a planning horizon sliding with each new update. User preferences are taken into account in the system design, and applied as criteria to select a single solution from the set returned by the multi-objective optimization procedure. To account for potential resource limitations facing the client site (deployed assets or command centres), a web-based decision support system where expensive computations are done remotely and where results are accessible through a thin client via the Internet is implemented. The resulting decision support system is demonstrated for a counter-piracy operation conducted off the coast of Somalia.
Keywords: Decision support system; Asset allocation; Counter-piracy; Environmental forecast; METOC; Optimization; Web service.