Michael James Horry
Michael graduated from the University of NSW in 1998 with a double degree in Engineering and Laws. Since that time he has held various industry roles and is currently an Executive Architect at IBM responsible for solution design across the partner and SI community. Michael is an Open Group Master Certified IT Architect as well as TOGAF 9 certified and a member of the Enterprise Architects Association of Australia.
Over his 14 years as an architect at IBM Michael has developed a reputation for finding creative solutions to problems helping customers to realise value by either reducing costs or creating opportunities for revenue with new or improved technical capabilities.
Michael developed an interest in Artificial Intelligence while working as IBM Services CTO for IoT – repeatedly using various Machine Learning methods to find patterns in big data sets sourced from IoT implementations. Michael commenced his PhD in 2020 to further pursue this interest.
Much of the recent research in the use of computer vision for medical imaging focuses on using Convolutional Neural Networks (CNNs) to identify and label abnormalities in medical images. Unfortunately, CNNs and other deep learning algorithms rely on huge corpuses of sensitised, labelled training data ideally from independent multiple sources. These methods also rely on high performance compute environment for training and to a lesser extent prediction.
Michael is investigating whether the use of hybrid AI models can address some of these shortcomings of CNNs. The goal of his research is to identify which combination of AI modelling techniques can provide acceptable performance on low performance compute environment thereby improving access to state-of-the-art computational diagnostic assistance to clinicians regardless of location or connectivity to the internet. Ideally this research will support a distributed model of AI whereby desensitized data can be accumulated in the field and shared with other clinicians for localized modelling and prediction.