Meeting Register Page

IC CAE Colloquium Series - The Generalizability Of Neural Networks: From Counting Crowds To Discovering Exoplanets
It's rare today to find a field of research wholly untouched by the subset of artificial intelligence known as neural networks. As data quantities grow, as computing power increases, and as modern neural network techniques mature, the influence of neural networks is certain to expand further. A primary element of the success of neural networks is their generalizability. This talk will follow the speaker's journey from developing neural networks for crowd analysis for assisting evacuation simulations to develop neural networks for discovering planets around other stars. Although vastly different fields of research, the underlying neural network techniques to solve each problem are near identical. In this talk, an intuitive understanding of how neural networks solve their tasks will be presented, with the hope that you will walk away having a sense of where neural networks may benefit your own area of research.

Greg Olmschenk is a postdoctoral fellow at NASA's Goddard Space Flight Center, where he develops artificial intelligence to identify various phenomena in astrophysical data. Prior research includes generative models, neural networks for infrastructure and security, and computer vision.

Sep 16, 2022 12:30 PM in Eastern Time (US and Canada)

* Required information