Predictive-Coding Neural Networks
The amazing progress of neural networks has given us clues about how the brain might perform some of the tasks that we consider to be intelligent: visual recognition, natural language understanding, strategic game play, etc. However, these architectures are quite different from the networks of neurons in your brain, so it’s still not clear how people manage to learn and perform these tasks. One promising direction for filling in this gap is the theory of predictive coding. I will describe a recent predictive coding approach by Rafal Bogacz, and show some preliminary results and current challenges in working with this architecture.
Jeff Orchard received degrees in applied mathematics from the University of Waterloo (BMath) and the University of British Columbia (MSc), and received his PhD in Computing Science from Simon Fraser University in 2003. Since then, he has been a faculty member in the Cheriton School of Computer Science at the University of Waterloo, and is currently the Director of the Centre for Computational Mathematics in Industry and Commerce. His main research focus is on computational neuroscience and artificial intelligence, using mathematical models and computer simulations of neural networks to understand how the brain works. He has also published research papers in image processing, medical imaging, and graphics.