Explainable and Robust AI from Case-Based Systems
A Case-based System solves problems by retrieving similar, previously solved problems from its memory, and reusing their solutions. The fundamental assumption is that ‘similar problems have similar solutions’. Explainable AI recognises the need for AI systems to be able to offer explanations to underpin their decisions, recommendations or predictions in contexts where transparency is important. Robust Intelligence should be able to cope with ill-defined domains and complex, changing realistic contexts, in contrast to the brittleness of many model-based systems. This talk will look at a number of case-based applications, and will investigate the contribution they make towards explainable and robust AI.
Susan Craw is a Professor of Computing at Robert Gordon University in Aberdeen where she has been Director of the IDEAS Research Institute, Head of Research for Design & Technology, and Head of Computing. She graduated with BSc Honours and MSc by Research in Pure Mathematics and a PhD in Computing Science from Aberdeen University. Her research in Artificial Intelligence developing innovative machine learning and data/text/web mining technologies for building intelligent systems is established over 30 years. Current research interests include Intelligent Information Systems, Case Based Reasoning, and Recommender Systems. In 2015 she was elected a senior member of AAAI in recognition of her significant contribution to the field of AI.