Speaker Biography
Artur d’Avila Garcez, FBCS, is Professor of Computer Science and Director of the Research Centre for Machine Learning at City, University of London. He holds a Ph.D. in Computing (2000) from Imperial College London. He co-authored two books: Neural-Symbolic Cognitive Reasoning (Springer, 2009) and Neural-Symbolic Learning Systems (Springer, 2002), and more than 150 peer-reviewed publications in Artificial Intelligence, Machine Learning, Neural Computation and Neural-Symbolic Computing. Garcez is president of the Neural-Symbolic Learning and Reasoning Association, and a member of the editorial boards and programme committees of many journals and international conferences. His research has received funding from the Nuffield foundation, the EU, IBM, CNPq and CAPES Brazil, the Daiwa Foundation, the Royal Society, Innovate UK, ESRC and EPSRC, UK.
Talk Abstract
In neural-symbolic computing, symbolic AI is combined with neural networks to achieve symbolic reasoning and explainable deep learning. In this talk, I will review the work on knowledge extraction from neural networks going back to the late 1990’s, including knowledge extraction: a sound approach, knowledge extraction from stacks of Restricted Boltzmann Machines, recurrent neural networks, and a recent application of knowledge extraction in the gambling industry to reduce harm from gambling. This real application shows the need for knowledge extraction not only from an explainable AI perspective, but also for system maintenance and performance improvement. Time permitting, I will also refer to very recent work on the extraction of probabilistic M of N rules from deep networks and the potential contributions of knowledge extraction to transfer learning and ethical AI.