PhD Defence | Biologically Plausible Deep Learning: Should Airplanes flap their Wings?
University of Amsterdam PhD candidate, Peter O'Connor will be defending his PhD thesis “Biologically Plausible Deep Learning: Should Airplanes flap their Wings?” on Wednesday 23rd September.
Peter O’Connor’s thesis examines how neural networks can effectively be trained while obeying biological constraints. O’Connor completed his research under the supervision of Max Welling and Efstratios Gavves (both from UvA).
Deep neural networks follow a pattern of connectivity that was loosely inspired by neurobiology. The existence of a layered architecture, with deeper neurons representing increasingly abstract features, was known from neuroscience long before it was used in machine learning. However, when one looks beyond superficial similarities, deep networks appear to be fundamentally different to their biological counterparts in the way they communicate, the way they learn, and the domain in which they operate.
O’Connor’s research is not only of academic interest. The brain, which by any estimate does vastly more computation than any existing computer, uses only about 20W of power – less than a light bulb. Understanding how it works may help us to build more efficient computing hardware. The results from this work help to see what a truly brain-like machine learning architecture may look like.