Information forensics and security

Multimedia signal processing and machine learning

Covers research in computer science and technical intelligence including machine intelligence for the advanced and automated management, as well as analyses of the growing volume of crime scene and evidence-related data embedded in various digital multimedia and communications forms.

Real-time identification of suspicious objects/persons in surveillance images and video

Surveillance images and videos or closed circuit television are commonly taken at important public areas such as airports and shopping centres. Currently, these videos are either monitored by security personnel who may not be able to continuously stay focused for extended period of time or stored for later review when some incidents have happened. Using such videos in either of these ways is ineffective for crime prevention or early crime detection.

This project aims to develop a number of techniques that can automatically monitor surveillance videos and alert security personnel when certain suspicious objects/persons/movements are identified. The developed techniques will have significant impact on improving public security. The supervisors have extensive research expertise in multimedia feature extraction, image registration and object tracking/recognition.

Proposed PhD supervisors: Dr Shyh Wei Teng (Principal), Prof Guojun Lu and Dr Dengsheng Zhang

Download scholarships application form (docx, 242kb)

Image/video annotation and retrieval using machine learning techniques for automatic evidence cataloguing

Current content-based image retrieval systems are mainly based on similarity of low level features. They are not ideal as humans recognise/retrieve images based on high concepts (e.g. objects and their relationships within images) and similar high-level concepts may have very different low-level features.

This project will develop a new image retrieval approach that will learn high level concepts using machine learning techniques and then annotate images automatically. The annotated images can then be organised and retrieved effectively and efficiently using text-based retrieval techniques. The developed system will have wide applications in areas (e.g. crime scene investigation) where a large number of images are used.

Proposed PhD supervisors: Prof Guojun Lu (Principal), Prof Kai Ming Ting, Dr Dengsheng Zhang and Dr Tanveer Choudhury

Download scholarships application form (docx, 242kb)

Multiview and free viewpoint video for digital crime scene reconstruction and analyses

Depth-based view synthesis in the multiview and free viewpoint video (MV/FVV) framework has gained recent research interest for superior image quality and coding efficiency. The main objective of this PhD research project is to develop a theoretical framework for 3D scene reconstruction and analysis using the video feeds from a MV/FVV camera setup with potential applications in crime scene investigation (CSI).

To validate the theoretical framework, the project also aims to develop cutting-edge software packages that use virtual tools augmented onto the 3D reconstructed crime scene to facilitate automated evidence searching, matching, and tagging (position, orientation, etc.). The project will be supervised by a team with exclusive research expertise in 3D video and augmented reality technologies, especially in image/video classification, segmentation, and coding. The student will have flexibility to work independently within this framework where external supervisors from the law enforcement agencies are expected to provide CSI-domain specific expertise.

Proposed PhD supervisors: Prof Manzur Murshed (Principal), Dr Shyh Wei Teng and Dr Gour Karmakar

Download scholarships application form (docx, 242kb)

Mass estimation as a means for data mining and machine learning with forensics applications

The primary aim of this project is to enable mining big data without the need for big data infrastructures by overcoming the underlying constraining weakness of current data mining approaches. It investigates the new enabling paradigm—mass estimation—with specific applications to forensics.

Proposed PhD supervisors: Prof Kai Ming Ting (Principal) and Sunil Aryal

Download scholarships application form (docx, 242kb)

Applying reinforcement learning to facilitate retrieval of information from a network of surveillance cameras

Law enforcement agencies are increasingly reliant on the use of networks of surveillance devices such as CCTV to rapidly identify crimes as they are occurring, or as a source of evidence for identifying, locating and/or prosecuting criminals. This project will develop intelligent software agents based on multiobjective reinforcement learning methods to predict and track a targeted individual’s direction through a network of smart cameras.

Proposed PhD supervisors: Assoc Prof Peter Vamplew (Principal) and Dr Richard Dazeley

Download scholarships application form (docx, 242kb)

Monitoring Internet-of-Thing objects in smart environment

Internet-of-Things (IoT) is increasingly attracting the attention of research community and industries worldwide as it is regarded as the main vehicle to make an environment smart. Because of this, it is burgeoning in every aspect of human life starting from advancing the global economy to improving the quality of life, productivity and security. The monitoring of IoT objects, which can be considered both beings and things, and tracing their behaviour and activities in the maze of cyber and physical worlds is a major research problem.

Currently, the research projects on this topic mainly consider the information derived from a physical system. This project aims to investigate the development of a framework for monitoring and tracing the abnormal behaviour of IoT objects using the footprints left in both cyber and physical worlds.

Proposed PhD supervisors: Assoc Prof Joarder Kamruzzaman (Principal), Dr Gour Karmakar, Prof Manzur Murshed and Dr Gayan Appuhamillage

Download scholarships application form (docx, 242kb)

Cyber security

Involves research in (i) cyber security, (ii) optimal and adaptive security, (iii) healthcare privacy, (iv) Internet of Things and software defined network security, and (v) analytics for cyber security attribution.

Big data approach to banking alerts, monitoring and decision making

Banks operate various information systems to provide services to their customers and these very systems are being targeted by the cyber criminals to gain unfair financial benefits. Integration of these systems creates vulnerabilities, which might be not there when the systems operate stand alone. Big data analytics techniques can be used to provide tailored services on the integrated systems with great efficiency to the customers. These systems have been monitored for cyber-attacks in past through their indivisible status logs, but for integrated systems, big data analytics approaches need to be investigated for dynamic alert analysis from heterogeneous sources. This project will investigate big data approaches for alert analysis from heterogeneous sources.

Supervisors: Assoc Prof Iqbal Gondal (Principal), Assoc Prof Joarder Kamruzzaman, Dr Mahmood Chadhar

Download scholarships application form (docx, 242kb)

Optimisation problems in real-time optimal adaptive security and approaches for their solution

Network security is about development of methods to prevent unauthorized access and misuse of computer networks including those in banks. Real-time adaptive security is the network security model necessary to reconfigure the network in a given time period so that to minimise possible losses and to prevent threats. The aim of this project to model real-time adaptive security problems using optimisation techniques, to develop algorithms for solving such problems and implement these algorithms in real life situations.

Supervisors: Assoc Prof Adil Baghirov (Principal), Assoc Prof Iqbal Gondal, Dr Julien Ugon

Download scholarships application form (docx, 242kb)

Ensuring patient privacy given Internet-of-Things health data transmitted from wearable sensors

Body area wireless sensor network (BAWSN) technology has a wide range of applications in healthcare, the military and sports as the sensor revolution known as the Internet-of-Things emerges. Ensuring that vital signs and related biological data generated with wearable sensors is processed and transmitted securely in a manner that protects the privacy of individuals while providing appropriate access to health care professionals, is a pressing challenge.

This project will develop an architecture that integrates access control mechanisms for the prevention of breaches with information accountability processes for the rapid detection of breaches if they occur.

Supervisors: Assoc Prof Andrew Stranieri (Principal), Dr Venki Balasubramanian, Assoc Prof Iqbal Gondal

Download scholarships application form (docx, 242kb)

Prudent assisted reinforcement learning for the identification of fraudulent user behaviour in online banking

Fraud detection in online banking is performed using tools such as Falcon and Proactive Risk Manager (PRM) which primarily use historical data to model potential frauds. These methods are unable to adapt at the speed fraudsters’ change their method of attack. A Prudent Assisted Reinforcement Learning (PARL) agent integrates machine learning methods to historical models allowing the agent to adapt to changing attacks rapidly. This project will adapt prudence based knowledge based systems (Maruatona, 2012) to be integrated with the Assisted Reinforcement Learning framework (Bignold, 2016) and applied to the detection of bank fraud.

Supervisors: Dr Richard Dazeley (Principal), Assoc Prof Peter Vamplew, Dr Rosemary Torney

Download scholarships application form (docx, 242kb)

Robust malware detection system for mobile platform

The number of mobile phone users is expected to reach 5 billion by 2017 and will continue to grow even at a faster rate. While this has contributed enormously to increased productivity, financial transaction and mobile commerce at one hand, on the other rising number of hackers are targeting mobile devices by generating and transmitting malicious contents. The fact that security of sensitive data on mobile devices is taken lightly further compounds the situation. Till now a robust mobile malware detection system is not available. This project targets to develop a robust malware detection and isolation system for mobile platform using computational intelligence and artificial immune systems, taking influencing factors into account, such as device constraints, user activity and profile, mobility, and operating system variations.

The scope of the project is not rigid and a prospective PhD student will have the flexibility to work within the greater scope of malware detection.

Supervisors: Assoc Prof Joarder Kamruzzaman (Principal), Assoc Prof Iqbal Gondal, Dr Cameron Foale

Download scholarships application form (docx, 242kb)

Efficient and real-time algorithms for cyber security data

Most available data in cyber security have a large number of documents and attributes and they are dynamic, changing rapidly. Development of efficient and accurate algorithms for dealing with this type of data is challenging as these algorithm should produce results in a given time period. The aim of this project is to develop new clustering algorithms which have high accuracy in detecting malware and malicious documents in real time.

Supervisors: Assoc Prof Adil Baghirov (Principal), Assoc Prof Iqbal Gondal

Download scholarships application form (docx, 242kb)