Dr David Cornforth

Biography

Dr David Cornforth is a Lecturer in Information Technology at the University of Newcastle. His research interests include data mining, artificial life and computer assisted medical diagnosis. He has conducted research in a range of areas – from synchronous and asynchronous processes in multi-agent systems, to intrusion detection systems for computer networks. Further, his foray into health informatics mainly pertains to visualisation in biomedicine, including image processing and analysis in neural networks, as well as retinal imaging.

He has successfully secured funding for his various projects including the Faculty Strategic Initiative Research Fund for his work in health informatics, and the New Staff Grant for his work in supervised and unsupervised data mining.

Dr Cornforth also brings with him the valuable expertise acquired during his time as a CSIRO research scientist in renewable energy. He currently teaches into the courses related to video games design, application programming and mobile apps development.

He is also a member of the School of Design, Communication and Information Technology’s research committee, as well as the Applied Informatics Research Group (AIR). Outside of academia, he is a member of the University of Newcastle Chamber Choir.

Qualifications

• PhD, University of Nottingham - UK, 1994
• Graduate Certificate in University Teach & Learn, University of New South Wales, 2007
• Bachelor of Science (Electrical & Electronic Eng), Trent Polytechnic, 1982

---

Research

Research keywords

• Applied Machine Intelligence
• Artificial Life
• Complex Systems
• Data Mining
• Electrical Networks
• Evolutionary Computation
• Health informatics
• Information Systems
• Multi agent systems
• Neural Networks
• Optimisation
• Pattern recognition

Research expertise

Current Research Projects:

• Computer assisted feature selection for supervised and unsupervised data mining

• Analysis of Heart Rate Variability using Multi Spectral Entropy Measures.

Heart rate can be measured with simple equipment and does not provide the full ECG signal. However, there is much information in the beat to beat (RR) interval. Heart rate variability has been analysed with time- and frequency-domain methods, but more recent nonlinear analysis has shown an increased sensitivity for identifying risk of future morbidity and mortality in cardiac patients.

• Finding patterns in large scale disturbances of the US electricity network.

A large electricity network can be analyzed and understood from a complex systems viewpoint. Power systems are believed to be examples of systems with Self Organizing Criticality. That is, such systems may evolve towards a critical point. The implications are that large blackouts are an inevitable feature of such networks.

Some of my research projects are:

• Models of social networks: This project investigated the impact of media on shared opinion within social groups, using network models. My contribution was in experimental design and analysis of results (Stocker et al., 2002a & 2002b; Stocker et al., 2003). This work showed that different network topologies (random, scale-free, small-world, and regular) affect the spread of ideas through a social network.

• Agent-based artificial markets: This project simulates a network of intelligent software agents trading with one another to form stable solutions. Agents are self-interested and have diverse goals, yet a network of peers with no designated leader is able to cooperate and find solutions. I observed that cooperation is easier to achieve in heterogeneous agents, because this allows tie breaking (Cornforth et al., 2004). I proposed and studied the use of an evolutionary algorithm to optimise the agents, providing better solutions in dynamic problems (Cornforth, 2007).

• Asynchrony in network models: I have researched different methods of asynchrony in multi agent models, implemented them and performed experiments using network measures such as degree distribution, clustering coefficient, state entropy and Lyapunov exponents. I have identified several types of updating schemes, and modeled the effects of these schemes using one-dimensional cellular automata. Results showed that the scheme chosen provides very different dynamics. I invented the self-synchronous scheme, where agents are coupled with their neighbours, so can update their state in synchrony with their neighbours. I showed that it is possible to switch between chaotic, cyclic and modular behaviour by varying a single parameter. The significance of this is a possible mechanism by which environmental parameters influence emergent structure. (Cornforth et al., 2005; Cornforth et al., 2002; Cornforth et al., 2001). Some of my software is online at VLAB, Monash University’s Complexity Virtual Lab, hosted at http://journal-ci.csse.monash.edu.au/vlab/ (e.g. Self Synchronous under Cellular Automata).

• Early detection of diabetes: Diabetes can be detected by advanced processing of retinal images. I have contributed measures that can be extracted from images and conducted experiments showing the high success rate of some pattern recognition algorithms (Cornforth & Jelinek, H.F. (2007, Cree et al., 2005a & 2005b & 2005c, Jelinek et al., 2005, Cornforth & Jelinek, 2005; Cornforth et al., 2004; Jelinek et al., 2004, Jelinek et al., 2002).

• Wrapper subset evaluation: This work aimed to determine the best set of measures for the automated classification of medical images. It is known that using too many measures has a detrimental effect upon the classifier performance. Although Principal Components Analysis is well established, it assumes a Gaussian distribution, whereas many modern classifier algorithms can model much more complicated and even concave dis

Collaboration

Charles Sturt University, NSW - Application of intelligent information systems to medical image and signal processing.

CSIRO - multi agent models of biological systems

Fields of Research

Code	Description	Percentage
080199	Artificial Intelligence And Image Processing Not Elsewhere Classified	50
080699	Information Systems Not Elsewhere Classified	30
080702	Health Informatics	20

Centres and Groups

Group

• Applied Informatics Research Group (AIR)

---

Administrative

Administrative expertise

Organizer of the AIR group

---

Teaching

Teaching keywords

• Games Design

Teaching expertise

I have a record of excellence in teaching, which is demonstrated by the results of student surveys. I am committed to improving the quality of the student experience, which led me complete a Graduate Certificate in University Learning and Teaching, awarded by UNSW in 2007. I have adopted the practice of keeping a teaching portfolio, where I record my reflection on my teaching practice. I have developed a culture of scholarship in teaching, demonstrated by innovations that have improved the learning experience for my students, based on the application of relevant literature on teaching and learning. Through introducing different teaching styles and techniques during lectures, I am committed to generating a richer student-centred learning environment.

My teaching experience includes:

• Computer Tools for Engineers
• Introduction to Computer Science
• Programming Fundamentals
• Introduction to Research in IT
• Knowledge Based Systems
• XML Technologies
• Soft Computing
• Research Methods
• Data Analysis for Postgraduates

I have had 10 years university teaching experience in three different universities and very diverse schools. I have been able to adapt to different topics, different course structure, and different modes of delivery. I have taught service courses for other schools in face-to-face mode and distance learning, with a highly diverse student cohort in multidisciplinary schools. I have transformed the teaching of my courses by introducing many student-centred exercises, and adapting the course to serve the needs of off-campus students.