Variational Graph Convolutional Neural Networks
This work addresses the need for uncertainty estimation in graph-based models for critical applications like social trading analysis and human action recognition, though it appears incremental as it adapts existing variational methods to graph convolutional networks.
The paper tackled the problem of improving explainability and accuracy in Graph Convolutional Networks by proposing Variational Neural Network versions for spatial and spatio-temporal graphs, which estimate uncertainty in outputs and attentions, resulting in improved model accuracy on datasets like Finnish board membership, NTU-60, NTU-120, and Kinetics.
Estimation of model uncertainty can help improve the explainability of Graph Convolutional Networks and the accuracy of the models at the same time. Uncertainty can also be used in critical applications to verify the results of the model by an expert or additional models. In this paper, we propose Variational Neural Network versions of spatial and spatio-temporal Graph Convolutional Networks. We estimate uncertainty in both outputs and layer-wise attentions of the models, which has the potential for improving model explainability. We showcase the benefits of these models in the social trading analysis and the skeleton-based human action recognition tasks on the Finnish board membership, NTU-60, NTU-120 and Kinetics datasets, where we show improvement in model accuracy in addition to estimated model uncertainties.