Gibbons Lectures - Deep learning - what's missing? Event as iCalendar

18 May 2017

6:30pm

Venue: 260.092 (OGGB3), Level 0

Location: Owen G Glenn Building, 12 Grafton Road, City Campus

Host: Department of Computer Science

MarcusFrean

Speaker:
Associate Professor Marcus Frean, School of Engineering & Computer Science, Victoria University of Wellington

There have certainly been some spectacular improvements in machine learning over the last couple of years and one has to wonder, what comes next? Associate Professor Frean will talk about recent breakthroughs but also focus on their intrinsic limitations in order to make some guesses about where the frontiers might lie. For example, the current paradigm of supervised learning is an important advance – but would unsupervised learning be more interesting if we could make it work? Neural nets have become much better at modeling some aspects of complex temporal data such as human language – but what about the aspects they're ill-disposed to learn? Traditional neural networks learn fixed mappings from inputs to outputs – what if they could learn to implement the actual algorithms themselves?

Marcus Frean studied Physics at Massey University and went on to complete a PhD in the Centre for Cognitive Science at Edinburgh University. Following a series of post-doc research positions at Cambridge, Otago and Queensland, he returned to New Zealand and joined Victoria University of Wellington in 2000. 

His primary research interest lies in building intelligent systems using machine learning involving inference and optimisation in neural nets, belief nets, and Gaussian processes. This is driven by the challenge of making computers learn by themselves and the continuing mystery of how real brains achieve the same thing. A second research direction is adaptive systems and evolution, in particular the emergent behaviour of complex communities. 

Drinks and nibbles will be served from 6pm on Level 1 of the Owen G Glenn building.

For more information on the Gibbons Lecture Series, click here.