Searching better, searching for more: helping HEP out of the new-physics crisis

Abstract: The standard model of particle physics is one of the pillars of our understanding of the fundamental laws of nature. Proposed in the ‘70s, this model passed 40 years of high-precision tests, culminating with the discovery of the Higgs boson by ATLAS and CMS in 2012. Despite this successful record, the standard model comes short in several ways, e.g. not describing gravity or not predicting dark matter. This is the reason why particle physicists are intensively looking for a discrepancy between theory predictions and experimental observations at colliders and in underground  experiments around the world. In this seminar, I will cover the last decade of searches for new physics at the LHC: what we learned from the LHC data, the limitations we faced and how we improved our techniques to overcome these limitations. With no signal of new physics found at the LHC or elsewhere, the HEP community is proposing a new generation of large-statistics experiments, to turn new physics searches into high-precision data analyses. Doing so with the current approach and the current online/offline computing technology will not scale. Even maintaining the current performances is not guaranteed. At the same time, one would need even more resources to extend the exploration into new directions. In this seminar, I will discuss how deep learning could be used as a tool to boost our physics reach in a cost-effective way, turning the HEP Big Data crisis into an opportunity for new research directions and, hopefully, new discoveries.