QCD-Aware Neural Networks for Jet Physics

<!--HTML--><p>Recent progress in applying machine learning for jet physics has been built upon an analogy&nbsp;between calorimeters and images. In this work, we present a novel class of recursive neural&nbsp;networks built instead upon an analogy between QCD and natural languages. In the analogy,&nbsp;four-momenta are like words and the clustering history of sequential recombination jet&nbsp;algorithms is like the parsing of a sentence. Our approach works directly with the four-momenta&nbsp;of a variable-length set of particles, and the jet-based neural network topology varies on an event-by-event basis. Our experiments highlight the flexibility of our method for building task-specific jet&nbsp;embeddings and show that recursive architectures are significantly more accurate and data&nbsp;efficient than previous image-based networks. We extend the analogy from individual jets&nbsp;(sentences) to full events (paragraphs), and show for the first time an event-level classifier&nbsp;operating on all the stable particles produced in an LHC event. I will discuss future directions for this style of hybrid physics-aware machine learning algorithms.</p>