GCNIllustrator: Illustrating the Effect of Hyperparameters on Graph Convolutional Networks

Ivona Najdenkoska, Jeroen den Boef, Thomas Schneider, Justo van der Werf, Reinier de Ridder, Fajar Fathurrahman, Marcel Worring

Abstract

An increasing number of real-world applications are using graph-structured datasets, imposing challenges to existing machine learning algorithms. Graph Convolutional Networks (GCNs) are deep learning models, specifically designed to operate on graphs. One of the most tedious steps in training GCNs is the choice of the hyperparameters, especially since they exhibit unique properties compared to other neural models. Not only machine learning beginners, but also experienced practitioners often have difficulties to properly tune their models. We hypothesize that having a tool that visualizes the effect of hyperparameters choice on the performance can accelerate the model development and improve the understanding of these black-box models. Additionally, observing clusters of certain nodes helps to empirically understand how a given prediction was made due to the feature propagation step of GCNs. Therefore, this demo introduces GCNIllustrator - a web-based visual analytics tool for illustrating the effect of hyperparameters on the predictions in a citations graph.
BibTeX
@inproceedings{najdenkoska2021gcnillustrator,
  author = {Ivona Najdenkoska and Jeroen den Boef and Thomas Schneider and Justo van der Werf and Reinier de Ridder and Fajar Fathurrahman and Marcel Worring},
  title = {{GCNIllustrator: Illustrating the Effect of Hyperparameters on Graph Convolutional Networks}},
  booktitle = {ACM Multimedia},
  year = {2021},
  doi = {10.1145/3474085.3478566},
}

Loading related content...

Figure