Semi-Equivariant GNN Architectures for Jet Tagging. Issue 1 (1st February 2023)

• Record Type: Journal Article
• Authors: Murnane, Daniel; Thais, Savannah; Wong, Jason
• Is Part Of: Journal of physics. Volume 2438:Issue 1(2023)
• Publication Date: 2023-02-01
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/2438/1/012121
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: Composing Graph Neural Networks (GNNs) of operations that respect physical symmetries has been suggested to give better model performance with a smaller number of learnable parameters. However, real-world applications, such as in high energy physics have not born this out. We present the novel architecture VecNet that combines both symmetry-respecting and unconstrained operations to study and tune the degree of physics-informed GNNs. We introduce a novel metric, the ant factor, to quantify the resource-efficiency of each configuration in the search-space. We find that a generalized architecture such as ours can deliver optimal performance in resource-constrained applications.