The APPLICATION STREAM will present papers describing practical applications of AI. This is the largest annual showcase in Europe of real applications using AI technology and the ideal place to meet with those working to make successful AI based applications.
As in previous conferences, the papers to be presented are expected to cover a wide variety of methods and application domains. The papers presented at recent conferences have included topics as diverse as automatic personalised TV listings, practical knowledge management, industrial planning, real time military decision support, financial fraud detection, improving manufacturing techniques and crime investigation, and featured the whole range of AI techniques, including CBR, fuzzy logic, rule based systems, neural networks, data mining and intelligent agents.
Papers will be selected to highlight critical areas of success (and failure) and to present the benefits and difficulties encountered. A key criterion in the selection of papers will be their relevance to other developers, to ensure that this stream is of value and interest to all those responsible for developing or managing practical AI based systems.
Invited Keynote Lecture
Dr Graham Ball, Nottingham Trent University, Nottingham, UK
Using Artificial Neural Network based computational algorithms for finding patterns in complex biomedical data.
The advent of gene expression micro-arrays has attracted considerable interest from biologists due to the potential for high throughput analysis of many thousands of genes. This technology has been applied to the classification of cancer patients. Subsequently, the data has been used to identify tissue expression profiles (molecular signatures) resulting in patient subgroups, characterised by unique phenotypes, corresponding with cancer associated disease pathways and clinical outcome. A potential handicap of such technologies is the unprecedented amount of data generated, often of high complexity and dimensionality. This has created a demand for novel data analysis methodologies. Here, we present a novel Artificial Neural Network (ANN) based methodology that accurately identifies a gene expression signature indicative of prognosis in breast cancer. By applying this approach to a previously published seminal data set we have identified 9 key genes whose expression profiles are capable of predicting development of distant metastases to median accuracies of 98%. Furthermore we present methods for the characterisation of the identified genes, development of population risk models and the study of interaction networks/pathways within the genes.
Dr Graham Ball is a reader in bioinformatics at the Nottingham Trent University. He has been involved in the development and validation of bioinformatic algorithms using Artificial Neural Networks for the last 16 years.
He is currently bioinformatics lead for 2 sixth framework projects one Strep (ENACT) and one Network of Excellence (BIOPATTERN) He is a UK representative for the DTI led UK-US biosciences partnerships focusing on cancer and infectious diseases. He was a UK representative for the United Nations working group on trans-boundary air pollution effects. He currently supervises a team comprising 5 members focusing on bioinformatic analysis of a range of domains.
His current research interests are directed at the application of computational bioinformatic techniques for the identification and validation of biomarkers that facilitate classification and characterisation of biological systems and populations. This is achieved through development of platform technologies centred on artificial neural network and other machine learning technologies that are applied to biological data (including proteomic, immunological and gene array sources). These platform technologies have been applied to the characterisation of patterns in microbial pathogen and cancer patient populations.