Why scholars are diagramming neural network models
This is a philosophical exploration that may inform better communication practices in AI research, but it is incremental as it synthesizes existing theories without proposing new methods or data.
The paper examines the diversity of diagrams used to communicate neural network architectures, analyzing how different diagrammatic choices reflect what aspects are prioritized in conveying these complex models.
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.