Automating Gravitational Lensing Detection
Time: 12:30 pm November 18th, 2015
Location: Building 53 Room 3004
Algorithms for detecting strong gravitational lensing candidates in current and future sky surveys must give rise to exceptionally small false positive rates, due to the high cost of validating lenses. This, combined with the fact that lenses are already rare, leads to low overall detection rates. I will talk about ongoing work to attack this problem using convolutional neural networks trained on images developed for the recent citizen scientist Space Warps project.