Year: 2020
Type: preprint
Abstract: Many problems in machine learning can be cast as learning functions from sets to graphs, or more generally to hypergraphs; in short, Set2Graph functions. Examples include clustering, learning vertex and edge features on graphs, and learning features on triplets in a collection. A natural approach for building Set2Graph models is to characterize all linear equivariant set-to-hypergraph layers and stack them with non-linear activations. This poses two challenges: (i) the expressive power of these networks is not well understood; and (ii) these models would suffer from high, often intractable computational and memory complexity, as their dimension grows exponentially. This paper advocates a family of neural network models for learning Set2Graph functions that is both practical and of maximal expressive power (universal), that is, can approximate arbitrary continuous Set2Graph functions over compact sets. Testing these models on different machine learning tasks, mainly an application to particle physics, we find them favorable to existing baselines.
Source: arXiv (Cornell University)
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Awards: NSF Student Travel Grant for the 2020 IFIP Networking Conference (IFIP NETWORKING), G3543283306, G4580243240, S2I2: Institute for Research and Innovation in Software for High Energy Physics (IRIS-HEP), Collaborative Research: Scalable CyberInfrastructure for Artificial Intelligence and Likelihood Free Inference (SCAILFIN), Collaborative Research: SI2-SSI: Data-Intensive Analysis for High Energy Physics (DIANA/HEP)
OpenAlex ID: W3008024694
