Tuesday 12pm, 30 October 2018
Deep learning with small education data: The unreasonable effectiveness of teacher priors.
Assistant Professor - Stanford University
The deep learning revolution promises that "with massive data and massive compute power we can train models to perform in highly intelligent ways on complex tasks", such as playing GO... or... perhaps understanding human learning. The only problem is: most education data is not "massive". Sad times. Will this deep learning revolution mostly be inapplicable to education? Luckily there is a solution. Come to the talk to learn more :-)
I was born and grew up in Nairobi, Kenya. When I was twelve I moved to Kuala Lumpur, Malaysia where I lived until I came to Stanford for university, liked it a lot and stayed. I am now an assistant professor in the CS department. My research lab focuses on machine learning to understand human learning. I love teaching and I'm into exploring our world (through both science and traveling).