Taylor Childers is a Computer Scientist at Argonne National Laboratory applying deep learning methods to scientific problems. He is interested in investigating how DL can replace historically developed analytical methods to speed up experimental science. Having begun as an experimental particle physicist at CERN, Taylor has been working with experimental data from the ATLAS experiment which measures the outcome of proton-proton collisions in the Large Hadron Collider.
The ATLAS detector is effectively a large camera (46m long, 25m diameter) containing more than 100 million detector pixels. Through many steps of data reduction and analysis, researchers convert the 100 million detector pixels into a handful of particles (electrons, photons, etc.). Taylor is studying how these traditional particle identification techniques, which take years to develop for each custom detector, can be replaced by deep networks.
Taylor got his PhD in Physics from the University of Minnesota (2007) studying cosmic ray with experiments flown on balloons above Antarctica. He then joined the ATLAS experiment as a post-doc with Heidelberg University (2007-2011) working on the custom hardware trigger electronics systems. This was followed by a CERN Fellowship (2011-2013) performing precision top quark pair cross-section measurements. He was hired into the ATLAS group at Argonne in 2013 and began work scaling collision simulations on Mira.