Physicists Learning from Machines Learning: Smart but Interpretable Neural Networks for Physics at the LHC50m
Machine Learning methods are extremely powerful but often function as black-box problem solvers, providing improved performance at the expense of clarity. Our work describes a new machine learning approach, which translates the strategy of a deep neural network into simple functions that are meaningful and intelligible to the physicist, without sacrificing performance improvements. We apply this approach to benchmark high-energy problems of fat-jet classification and electron identification. In each case, we find simple new observables that provide additional classification power and novel insights into the nature of the problem.
Taylor Faucett is a graduate student at the University of California, Irvine where he works on research applications of big data techniques and machine learning to topics in high-energy physics. In particular, his interests lie in the design of novel machine learning architectures which transform complex black-box neural networks into interpretable models which can yield new insights into physics studied at the LHC. Currently, Taylor is studying as a Chateaubriand Fellow at the Université Clermont Auvergne (UCA) in Clermont-Ferrand, France.