Graph Neural Networks and Time Series as Directed Graphs for Quality Recognition
This work addresses quality recognition problems, but appears incremental as it applies existing GNN methods to a new representation of time series.
The authors tackled quality recognition by modeling time series as directed graphs to encode time dependencies, and developed two geometric deep learning models (a classifier and an autoencoder) that achieved unspecified results.
Graph Neural Networks (GNNs) are becoming central in the study of time series, coupled with existing algorithms as Temporal Convolutional Networks and Recurrent Neural Networks. In this paper, we see time series themselves as directed graphs, so that their topology encodes time dependencies and we start to explore the effectiveness of GNNs architectures on them. We develop two distinct Geometric Deep Learning models, a supervised classifier and an autoencoder-like model for signal reconstruction. We apply these models on a quality recognition problem.