David Shih (Rutgers/LBNL) “Searching for the Unexpected at the LHC with Unsupervised Machine Learning”

ABSTRACT:

Conventional methods for searching for new physics at the LHC have mostly been “top-down”: starting from a specific model, searches are designed and optimized to have the best sensitivity to that model. Despite hundreds of conventional new physics searches at the LHC, none have turned up any hint of new physics. Maybe it’s time to admit that we don’t know what we’re looking for.

Breakthroughs in modern deep learning have the potential to revolutionize how we search for new physics at the LHC. In particular, techniques borrowed from unsupervised machine learning could enable us to search for new physics in a largely model-agnostic way. In this talk I will review some promising recent proposals in this direction. These proposed search strategies could complement more conventional methods by finding surprising signals that were not anticipated by any model, ensuring that we leave no stone unturned in the hunt for new physics at the LHC.