CapsNets continuing the convolutional quest
Sascha Diefenbacher, Hermann Frost, Gregor Kasieczka, Tilman Plehn, Jennifer Thompson
Abstract
Capsule networks are ideal tools to combine event-level and subjet information at the LHC. After benchmarking our capsule network against standard convolutional networks, we show how multi-class capsules extract a resonance decaying to top quarks from both, QCD di-jet and the top continuum backgrounds. We then show how its results can be easily interpreted. Finally, we use associated top-Higgs production to demonstrate that capsule networks can work on overlaying images to go beyond calorimeter information.
Topics & Concepts
Computer scienceBenchmarkingArtificial intelligenceIdeal (ethics)Convolutional neural networkResonance (particle physics)Production (economics)Natural language processingTheoretical computer scienceKey (lock)OverlayPhysicsAlgorithmOrder (exchange)QuarkPruningPattern recognition (psychology)Particle physics theoretical and experimental studiesQuantum Chromodynamics and Particle InteractionsHigh-Energy Particle Collisions Research