Interdisciplinary Dean's Speaker Series in Data Science: Andrew Gordon Wilson

Day Friday, November 08
Time 3:30 PM to 5:00 PM
Where LH-10


Andrew Gordon Wilson, assistant professor at the Courant Institute of Mathematical Sciences and Center for Data Science at New York University, will give a talk titled "How do we build models that learn and generalize?" 
A reception will follow at 4:30 p.m. in CW-112. 

RSVP at http://bit.ly/DS-TAE-RSVP.

To answer scientific questions, and reason about data, we must build models and perform inference within those models. But how should we approach model construction and inference to make the most successful predictions? How do we represent uncertainty and prior knowledge? How flexible should our models be? Should we use a single model, or multiple different models? Should we follow a different procedure depending on how much data are available?

In this talk, he will present a philosophy for model construction, grounded in probability theory. He will exemplify this approach for scalable kernel learning and Gaussian processes, Bayesian deep learning, and understanding human learning.


His webpage is https://cims.nyu.edu/~andrewgw.


Add to Calendar 08/11/19 3:30 PM 08/11/19 5:00 PM 15 Interdisciplinary Dean's Speaker Series in Data Science: Andrew Gordon Wilson <br>Andrew Gordon Wilson, assistant professor at the Courant Institute of Mathematical Sciences and Center for Data Science at New York University, will give a talk titled "How do we build models that learn and generalize?"&nbsp;<br>A reception will follow at 4:30 p.m. in CW-112.&nbsp;<br><br>RSVP at <a href="http://bit.ly/DS-TAE-RSVP">http://bit.ly/DS-TAE-RSVP</a>.<br><br>To answer scientific questions, and reason about data, we must build models and perform inference within those models. But how should we approach model construction and inference to make the most successful predictions? How do we represent uncertainty and prior knowledge? How flexible should our models be? Should we use a single model, or multiple different models? Should we follow a different procedure depending on how much data are available?<br><br>In this talk, he will present a philosophy for model construction, grounded in probability theory. He will exemplify this a LH-10 DD/MM/YY