Why scholars are diagramming neural network models

TL;DR

This paper explores how diagrams are used to represent neural network models, analyzing their diversity to understand what aspects of the models are prioritized for communication.

Contribution

It provides a philosophical analysis integrating theories of conceptual models, communication, and semiotics to interpret neural network diagrams.

Findings

Diagrams vary widely in representing neural networks.

Diversity reflects different communication priorities.

Semiotic analysis reveals underlying conceptual choices.

Abstract

Complex models, such as neural networks (NNs), are comprised of many interrelated components. In order to represent these models, eliciting and characterising the relations between components is essential. Perhaps because of this, diagrams, as "icons of relation", are a prevalent medium for signifying complex models. Diagrams used to communicate NN architectures are currently extremely varied. The diversity in diagrammatic choices provides an opportunity to gain insight into the aspects which are being prioritised for communication. In this philosophical exploration of NN diagrams, we integrate theories of conceptual models, communication theory, and semiotics.