Master of Research: Proposed Projects

You can select your project by one of the following methods:

  1. choosing a project that’s already been proposed by an academic (see below);
  2. going and talking to an academic in an area of research you’re interested in that the Department engages in, and arriving at a project topic from there; or
  3. taking a project proposal you’ve devised to an academic and convincing them that they should supervise it.

For option #1, below are some projects that have been proposed by academics in the Department of Computing.

Centre for Advanced Computing – Algorithms and Cryptography (ACAC)

Project Title: Analysing information flows in cryptographic APIs

Supervisor: A/Prof Annabelle McIver
Brief Description: In many security applications, controlling secret information is critically important. Unfortunately information can be accidently released in the operation of software (think of the notorious “side-channel” attacks). This project focusses on studying how information can potentially be released by cryptographic APIs (see e.g. http://www.bouncycastle.org/) with the goal of improving their specification, as well as determining how to improve testing and implementation.
Prerequisites:

Project Title: Combining data mining with security and privacy
SupervisorA/Prof Annabelle McIver
Brief Description: In many applications secret information can be accidently released with the result that a user’s privacy becomes compromised.  In badly-designed software, for example such information can compromise users’ privacy. This project aims to combine data mining methods with new techniques for measuring the severity of attacks in security systems. Applications include privacy in health informatics and electronic voting.
Prerequisites: A knowledge of datamining or machine learning techniques.

Project Title: Information propagation in vehicle-to-vehicle mobile wireless networks
Supervisor: Prof Bernard Mans
Brief Description:
Prerequisites:

Project Title: Information propagation in handheld-to-handheld mobile wireless networks
Supervisor: Prof Bernard Mans
Brief Description:
Prerequisites:

Project Title: Information propagation in time-varying graphs
Supervisor: Prof Bernard Mans
Brief Description:
Prerequisites:

Project Title: Exploration in time-varying graphs
Supervisor: Prof Bernard Mans
Brief Description:
Prerequisites:I need the information by mid

Project Title: Decentralised monitoring and alarming systems: algorithms and design
Supervisor: Prof Bernard Mans
Brief Description:
Prerequisites:

Project Title: Colouring in Cayley graphs
Supervisor: Prof Bernard Mans
Brief Description:
Prerequisites:

Project Title: Algorithmic problems in Cayley graphs
Supervisor: Prof Bernard Mans
Brief Description:
Prerequisites:

Project Title:Efficient, automated image analysis of large astronomical datasets
Supervisor: Prof Bernard Mans, Dr. Lee Spitler (Dept. of Physics & Astronomy)
Brief Description: This data scientist-like project focuses on the development of new algorithms to efficiently extract information from large (~Tb) astronomical image datasets. The challenge with astronomical imaging is to isolate the signal from extremely faint astronomical sources. Traditionally, this required custom-built analytical tools for each dataset. For many new astronomical telescopes, the rates of data collection are so high that it is no longer feasible to develop tools for each individual dataset. New algorithms are needed to efficiently process and analyze large datasets in order to extract the faint signals from astronomical sources. This MRes or PhD project will focus on new data being acquired as a part of the Macquarie-led Huntsman project (https://www.facebook.com/HuntsmanEye), which makes use of off-the-shelf Canon lenses to explore ultra-faint regions around nearby galaxies.
Prerequisites: None.

Advanced Cyber Security Research Centre (ACSRC)

Project Title: Security and Software Defined Networking
Supervisor: Prof Vijay Varadharajan
Brief Description: Software Defined Networking (SDN) is a set of technologies for allowing greater control over the operation of networks. SDN allows for optimising and configuring the network functions in a dynamic manner rather than a fairly static network that can only be controlled by proprietary vendor specific protocols, with sometimes limited visibility into the internals of network devices such as routers and switches. Furthermore, SDN can be achieved using commodity server hardware, which can add to the practicality and cost savings. This flexibility is network design is in part accomplished by separating the switch’s control plane from the data plane. Having this new level of control on the one hand can be of great benefit while on the other hand can lead to security vulnerabilities and pitfalls. The PhD is about the study of security properties in SDN and design of secure applications for managing SDN. The work will start with the study of OpenFlow SDN architecture.
Prerequisites:

Project Title: Security in Smart Grid Infrastructures
Supervisor: Prof Vijay Varadharajan
Brief Description: A critical area where cyber security plays a vital role is in the transformation of nation’s aging power grid into advanced smart grid infrastructure. A key issue is devising effective strategies for securing computing and communication networks that are central to the availability and performance of the energy infrastructure and for protecting the privacy of data associated with this smart infrastructure. PhD work will address security and privacy issue sin smart grid infrastructures. It will investigate the types of attacks that are possible in this emerging infrastructure. It will specify security and privacy policies for home based networks and multiple network devices. It will develop new techniques for addressing privacy of personal information in smart grid.
Prerequisites:

Project Title: Security and Big Data
Supervisor: Prof Vijay Varadharajan
Brief Description: Heterogeneity, scale, timeliness, complexity, and security and privacy problems associated with Big Data are now being commonly recognised. In general, there are multiple stages associated with Big Data such as acquisition, extraction, modelling and analysis. Each of these stages introduces security and privacy problems in a distributed environment. PhD in this area could study one or more of these problems. Let me give just an example. Data provenance is often a critical issue when it comes to decision making. The meta data associated with the history of data is needed in determining the integrity and trustworthiness of data. The work could start with analysing various approaches to analysing provenance in cloud and how they can be improved to enhance the quality of decision making.
Prerequisites:

Project Title: Security and Cognitive Radio
Supervisor: Prof Vijay Varadharajan
Brief Description: A critical challenge is that spectrum is a finite resource and it is in great demand. Cognitive radio aims to address the limited availability of spectrum to cope with the proliferation of wireless services. Cognitive radio networks are invaluable in the elimination of this artificial shortage by providing the flexibility to users to access licensed parts of the spectrum. However there must be economic incentives must be in place to incentivise the license holders (primaries) to use the spectrum they have licensed in an intelligent manner, and thereby facilitate access by the rest (secondaries). For example, license-holders should be allowed to sell their white spaces (unused spectrum bands) in an open spectrum market, which needs to be designed while remaining cognizant of the features that distinguish spectrum trade from that of any other commodity (e.g. one distinctive feature that transmissions at neighbouring locations on the same channel interfere with each other, whereas the same channel can be used at far-off locations without mutual interference).
The project will investigate the following aspects: (1) Secure Access to Spectrum, and (2) Secure and fair protocols for dynamic spectrum transactions.
Prerequisites:

Project Title: Privacy negotiation between users and service providers
Supervisor: Dr Michael Hitchens
Brief Description: Privacy is an increasingly important consideration in online applications. For example, when a user wishes to access a resource the user wishes to reveal the minimum amount of information about themselves to access the resource. A service provider might have a number of policies governing the use of a resource, to preserve its own privacy, only wishes to reveal the minimum information about these policies to the user. This project will examine how users and service providers can negotiate, so that users can obtain access to resources while both parties maintain the maximal possible privacy.
Prerequisites:

Centre for Language Technology (CLT)

Project Title: Natural language data mining of social media
Supervisor: Prof Mark Johnson
Brief Description: This project involves applying data mining methods to a large collection of social media messages (probably Twitter) to study the relationship between the messages and the events they describe. For example, you could use machine learning methods to find the words or phrases in tweets about financial topics that are strong indicators of stock price movements or other financial measures. Then you’ll use these words or phrases as features to study the temporal or spatial distribution of information about different kinds of events. This has applications in financial markets, including stock price prediction and financial fraud detection.
Prerequisites:

Project Title: A comparative evaluation of on-line learning algorithms for data mining
SupervisorProf Mark Johnson
Brief Description: There’s been an explosion of on-line algorithms in the past decade for training data mining classification procedures. This project will compare several state-of-the-art algorithms on several different data sets in order to identify what kinds of data each algorithm does best on.
Prerequisites: Programming ability, matrix linear algebra and calculus.

Project Title: Integrating qualitative and quantitative information in data mining
SupervisorProf Mark Johnson
Brief Description: While data mining techniques can theoretically exploit both quantitative and qualitative information simultaneously, there are practical issues (e.g., involving scaling) that often need to be overcome.  This project involves identifying one or more data sets that involve both quantitative and qualitative information (e.g., financial data), applying several machine learning methods to that data and evaluating how well those methods exploit both the quantitative and qualitative information in the data.
Prerequisites: Programming ability, matrix linear algebra and calculus.

Project Title: Optimisation Techniques for Query-based Summarisation of Clinical Publications
Supervisor: Dr Diego Mollá Aliod
Brief Description: We have a corpus of clinical questions and answers. Each answer has a list of publication references arranged in groups according to how they address the answer. For example, if the question is “what is the best treatment for X”, and there are three possible treatments, then each treatment will have a group of references associated to it. The goal of this project is to use optimisation techniques in general, and Integer Linear Programming (ILP) in particular, to automatically summarise the texts of the groups of abstracts. For more background of the kind of research that I am doing, look at http://web.science.mq.edu.au/~diego/medicalnlp/.
Prerequisites: Knowledge of statistical methods and natural language processing, or willingness to learn these. Good programming skills, preferably using the Python programming language and Python packages such as NLTK and Numpy/Scipy. Experience with Integer Linear Programming (ILP).

Project Title: Topic Modelling for Query-based Summarisation of Clinical Publications
Supervisor: Dr Diego Mollá Aliod
Brief Description: We have a corpus of clinical questions and answers. Each answer has a list of publication references arranged in groups according to how they address the answer. For example, if the question is “what is the best treatment for X”, and there are three possible treatments, then each treatment will have a group of references associated to it. The goal of this project is to use topic modelling in general, and variants of Latent Dirichlet Allocation (LDA) in particular, to automatically summarise the texts of the groups of abstracts. For more background of the kind of research that I am doing, look at http://web.science.mq.edu.au/~diego/medicalnlp/.
Prerequisites: Knowledge of statistical methods and natural language processing, or willingness to learn these. Good programming skills, preferably using the Python programming language and Python packages such as NLTK and Numpy/Scipy.

Project Title: Find Relevant Medical Publications
Supervisor: Dr Diego Mollá Aliod
Brief Description: The goal of this project is to design and apply search technology so that a search retrieves published medical information that is relevant to the query asked by a medical doctor. Examples of medical queries are the titles of the clinical inquiries column of the Journal of Family Practice. You will use these clinical inquiries as your source questions, and you will aim to retrieve documents like those listed in the clinical inquiries, by accessing the MEDLINE database of medical publications through interfaces such as PubMed. For more background of the kind of research that I am doing, look at http://web.science.mq.edu.au/~diego/medicalnlp/.
Prerequisites: You will preferably have knowledge and experience with using Web applications, such as search services. Knowledge of search technology and text processing is also desirable but not essential. Good programming skills, preferably using the Python programming language and Python packages such as NLTK and Numpy/Scipy.

Project Title: Linking Language Resources with DBPedia
Supervisor: A/Prof Steve Cassidy
Brief Description: Language resources are collections of text, video and audio along with annotations that describe their structure. The HCS vLab project has built a large repository of LRs and an associated API and tool collection with the goal of supporting language and communication research. At the moment, all of the annotation is syntactic but there are a lot of tools available that will perform semantic analysis, for example linking names to DBPedia entries. The aim of this project is to explore these tools in the context of supporting linguistic enquiry. You will exploit the capabilities of the current web platform and explore how you might build better search and browsing tools for LRs using the augmentations possible with these semantic analysis technologies.
Prerequisites: This project will involve web expertise but applies this to look at supporting research methods in language and communication. You will probably build a neat web application and then evaluate it with the target audience.

Project Title: Evaluation and Optimisation of Annotation Query Systems
Supervisor: A/Prof Steve Cassidy
Brief Description: The HCS vLab is a large scale web platform for storing language data (text, audio, video) along with annotations and meta-data represented as RDF. While it provides some basic query services there is a lot of scope for building more sophisticated query systems on the basic service. Query of annotations has a long history and there is some work on describing the expressive power of query languages for particular problems.

This project would look at developing a standardised benchmark for evaluating query languages for annotation data based on the annotation store in HCS vLab. It would survey the existing query languages and implementations and try to implement one or more of these against the annotation store. It may also then look to implementing a more general query sub-system or optimising some aspect of one of the existing query languages.

A goal of the project would be to provide a benchmark platform that other researchers could use to evaluate new ideas in annotation query.
Prerequisites:

Project Title: Answer Set Programming for the Semantic Web
Supervisor: Dr. Rolf Schwitter
Brief Description: Answer Set Programming is a novel form of declarative programming that has its roots in logic programming, non-monotonic reasoning and knowledge representation. Existing rule systems for the Semantic Web fall into three categories: first-order, logic programming, and action rules. In this project, you will investigate the suitability of Answer Set Programming as a rule language for the Semantic Web and compare it with other rule paradigms.
Prerequisites:

Project Title: Friends and Foes in the Eurovision Song Contest
Supervisor: Dr. Rolf Schwitter
Brief Description: The Eurovision Song Contest is a yearly song competition held among the members of the European Broadcast Union since 1956. This contest involves the live television broadcast of popular songs from various European countries. Each country votes for its favourite songs based on a combination of televoting and jury. Over the years, it has been speculated that some complex tactical voting is going on between countries that this voting process disadvantages the participants of some countries and favours others. Using data from previous contests, you will use machine learning techniques to analyse the underlying voting mechanics of the Eurovison Song Contest and test the hypothesis of tactical voting.
Prerequisites:

Project Title: Business Rules for Humans and Machines
Supervisor: Dr. Rolf Schwitter
Brief Description: Business rules are statements that describe how an organisation should operate. Since business rules are usually written in a natural language such as English, they are often ambiguous and vague, and therefore difficult to process by a machine. This situation leads to suboptimal decisions making in organisations since the relevant knowledge is only available informally and (in various shapes and interpretations) in people’s heads. In this project, you will investigate how business rules can be reconstructed in a controlled natural language so that these rules are both human-readable and machine-processable for consistency checking.
Prerequisites:

Computational Epidemiology & Evidence Surveillance Lab

[Note that these proposed projects are from the Australian Institute of Health Innovation at Macquarie, but with joint supervision by academics in Computing.]

Project Title: Geo-locating Twitter users without location information using social network structure information
Supervisors: Dr Adam Dunn, Prof Mark Johnson
Brief Description: In order to do useful public health surveillance on Twitter, it is necessary to be able to locate individuals even when that information is not clearly available. Very recently, algorithms have been proposed that take advantage of information from the content of tweets, as well as information from the social network structure built from followers and mentions. In this project, you will develop and evaluate new algorithms designed for this task, and contribute directly to new public health surveillance methods developed in the Centre for Health Informatics.
Prerequisites: Knowledge of data mining techniques and good programming skills (Python/Matlab) are recommended. Knowledge of graph theory and familiarity with the Twitter APIs and GIS would be valuable. This is a challenging project.

Project Title: Automatic query expansion in topic-driven searches over Twitter
Supervisors: Dr Adam Dunn, Prof Mark Johnson
Brief Description: Surveillance of public health topics on Twitter relies critically on constructing a reliable set of search terms to capture any tweet that fits the topic but patterns of language change over time. In this project, you will use the Twitter API to investigate and evaluate query expansion methods for applications in public health, and contribute directly to new public health surveillance methods developed in the Centre for Health Informatics.
Prerequisites: A strong background in programming (Python/Matlab) and statistics. Experience in machine learning and the Twitter API would be desirable. The ability to work in close collaboration with team members is important here.

Project Title: Anti-vaccine sentiment analysis of webpages using unsupervised machine learning methods
Supervisors: Dr Adam Dunn, Prof Mark Johnson
Brief Description: Anti-vaccine groups and celebrities are actively engaged in subverting public health interventions by promoting misinformation about the safety and efficacy of vaccines. On social media, links to webpages about vaccines largely include peer-reviewed literature, mainstream media news, blogs, and YouTube videos. Information on these pages can help us to classify users and tweets, track the spread of information through specific communities. In this project, you will apply state-of-the-art methods in sentiment analysis to classify webpages linked from Twitter
Prerequisites: Strong programming skills (Python/Matlab) and some experience in machine learning. Experience in web-crawling and data mining (or data structures, databases, and algorithms) would be valuable.

Project Title: Using machine learning to automatically link registrations and publications of clinical trials
Supervisors: Dr Adam Dunn, Dr Diego Mollá Aliod
Brief Description: Clinical trials make up a large part of the evidence that doctors and their patients need to be able to make good decisions about their care. However, the results of clinical trials are not always published and they are not always reported completely when they are published. To help us to create automatic methods for determining if and when clinical trials are published and how completely they are reported, you will develop a system for monitoring new published trials on PubMed and find the matching registration on ClinicalTrials.gov if it exists. How you do this is up to you – you could choose to use machine learning, network algorithms, or purely statistical methods.
Prerequisites: Experience in data mining (or data structures, databases, and algorithms), programming (Python/Matlab), and statistics would be essential. This is a challenging project.

Project Title: Author name disambiguation on PubMed by combining text-mining and network structure information
Supervisors: Dr Adam Dunn, Dr Diego Mollá Aliod
Brief Description: The deceptively complicated problem of finding all the articles written by a single author can be translated into a discrete mapping problem. Your aim in this project will be to combine the information from co-authorship networks and text-based information from article abstracts and meta-data to develop and evaluate new forms of author disambiguation. The methods you develop will contribute directly to new forms of evidence surveillance that track the biases in medical literature that distort the consensus of safety and efficacy evidence for clinical interventions.
Prerequisites: A strong interest in algorithm development and data mining skills (or data structures, databases, and algorithms) are recommended.

Project Title: Predicting citations from new publications to old publications using supervised machine learning
Supervisors: Dr Adam Dunn, Dr Diego Mollá Aliod
Brief Description: New articles on known topics might be expected to cite key articles covering the same topic – when they do not it might represent a problem. When many new articles fail to cite key articles, this provides evidence of selective citation bias and possible problems in the evidence base. Your aim in this project will be to construct statistical models or use machine learning methods to predict citations from new articles to identify “missing” citations. Your new tools could be used to help recommend citations based on preliminary citation lists, or to track down research biases that distort the consensus of safety and efficacy for clinical interventions.
Prerequisites: Experience in machine learning and statistics would be essential here. This is a challenging project requiring a strong applied mathematics background and programming experience (Python/Matlab). An interest in applied machine learning methods would be beneficial.

Project Title: Measuring gap in appropriateness of care for people living with bipolar disorder or schizophrenia
Supervisors: Dr Oscar Perez-Concha, Prof Mark Johnson / Dr Diego Mollá Aliod
Brief Description: People living with bipolar disorder or schizophrenia have a life expectancy of 11 to 16 years lower than the general population, higher rates of medical comorbidities and poorer physical health outcomes. We think this might be because these people do not receive the same quality of preventative care for other conditions.
In this project you will use data mining methods to analyse hospital data to find out if people living with bipolar disorder or schizophrenia have a higher number of potentially preventable hospitalisations (i.e. due to diabetes complications that should have been treated in primary care) and higher mortality rates following these preventable hospitalisations.
Prerequisites: Knowledge of data mining techniques; good statistical and programming skills (R) are recommended.

Project Title: Hunting software robots: trawling the internet for scientific papers
Supervisors: Dr Diego Mollá Aliod, Guy Tsafnat
Research Area: Artificial Intelligence, Pattern Recognition, Information retrieval
Brief Description: Obtaining the full text of scientific papers is a tedious and time consuming but necessary task for many professionals including scientists and clinicians. While titles and abstracts can be found in search engines, the full text could be hidden on institutional web sites, behind pay-walls or on a personal web site not indexed by citations databases. Trawling the web for open versions, or emailing authors automatically are alternative methods to obtain the full text automatically. This project is to develop the artificial intelligence needed to hunt for scientific papers from their citations.
Prerequisites: Experience with web search and information retrieval, or willingness to learn these. Ideally, experience with text mining. Good programming skills and experience with Web applications.  Option for multiple students to work on aspects of the project in parallel.

Project Title: Extraction of PICO elements from the clinical trial papers
Supervisors: Dr Diego Mollá Aliod, Guy Tsafnat
Brief Description: Clinical trials are the cornerstone of evidence based medicine as they serve as the key source of evidence. To support obtaining key information elements from trials the PICO system has been developed. According to it, important information in clinical trials are the Problem (e.g. disease) that the trial studies, the Intervention (e.g. drug) used, the Comparison (e.g. alternative drug) against which the trial is done, and the Outcome (e.g. number of days in hospital) that is being used in the comparison. Many automatic tools have been proposed to automatically extract PICO elements from scientific papers that report on the findings of clinical trials. This project is to survey, and compare algorithms that extract PICO elements and to build a unified tool that combines the best algorithms.
Prerequisites: Experience with statistical methods and machine learning, or willingness to learn these. Good programming skills.

Intelligent Systems Group (ISG)

Project Title: Uncovering missing pieces in history through simulations in virtual worlds
Supervisor: Prof Deborah Richards
Brief Description: This project will involve creating simulations of an ancient Greek port at different time periods using current data and predictive models. These simulations will be used to understand changes in the landscape and uses of the port due to weather, cultural, geological, human society and other influences over a number of millennia. This is a multi-disciplinary project that will involve working with the ancient history department.
Prerequisites: Good programming skills are essential with aptitude to gain computer graphics and virtual world programming skills.

Project Title: Empathic agents for health and well-being
Supervisor: Prof Deborah Richards
Brief Description: This project will explore a topic involving the use of intelligent virtual agents to build a therapeutic alliance with the patient to deliver improved health outcomes. The project will involve collaboration with peadiatric specialists at the Children’s Hospital Westmead.
Prerequisites: Good programming skills.

Project Title: Adaptive Collaborative Learning
Supervisor: Prof Deborah Richards
Brief Description: This project aims to explore how data captured while using a virtual world for educational purposes can be used to scaffold the learning experience, assess the student’s progress and possibly inform the teacher of progress or issues.
Prerequisites: The student must have good programming skills. Ideally the student will have some experience with language processing and Unity3D.

Project Title: (Semi-)Automated Coding of Qualitative Responses
Supervisor: Prof Deborah Richards
Brief Description: The project aims to recommend, and ideally validate, appropriate approaches to (semi-)automatically process qualitative responses collected during experimental studies.
Prerequisites: The student must have good communication and writing skills. Ideally the student will be familiar with natural language processing (NLP) approaches. Programming skills to implement and validate recommended approaches is highly desirable.

Project Title: Uncovering missing pieces in history through simulations in virtual worlds
Supervisor: Prof Deborah Richards
Brief Description:
Prerequisites:

Project Title: Improving compliance with online advice for medical treatments
Supervisor: Prof Deborah Richards
Brief Description:
Prerequisites:

Project Title: Overlaying Business Processes with Social Networks
Supervisor: Dr Peter Busch
Brief Description: The goal is to refine techniques for ‘overlaying’ business processes with social networks as a means of determining the fit of tasks to people – comparing management viewpoints with actual work processes. Such research enhances workplace efficacy significantly. Tools and Techniques to be used include Business Process Management and Modelling, Petri Nets and Social Network Analysis graphs.
Prerequisites:

Programming Languages Research Group (PLRG)

The PLRG has around 15 project proposals described on their website.

Virtual and Interactive Simulations of Reality (VISOR)

Project Title: Stereoscopic visualisation and fragment shaders for the correction of optical defects such as astigmatism
Supervisor: A/Prof Manolya Kavakli
Brief Description: Auto-stereoscopic visualisation emerges as one of the most innovative technologies for the future of 3D displays such as Oculus Rift and Google glasses. In this thesis we focus on the generation of virtual and augmented 3D content using game engines in virtual reality and augmented reality settings. To produce realistic effects in 3D scenes, game engines usually make use of the GPU programming advances, such as shaders. Shaders allow the programmers to modify the graphics pipeline, providing the possibility of programming new actions for the vertices, fragments or geometry of the scene. This is performed directly on the GPU, generating very fast operations that are executed in parallel with the operations executed on the CPU. Our goal in this thesis is to test the performance and usability of fragment shaders in the correction of optical defects such as astigmatism. Testing with several participants will be done in order to analyse the visual results and draw comparisons between virtual reality and augmented reality settings using a number of shaders. We will develop the system in a game engine, such as Unity 3D.
Prerequisites: