Directed Graph Auto-Encoders
Abstract
We introduce a new class of auto-encoders for directed graphs, motivated by a direct extension of the Weisfeiler-Leman algorithm to pairs of node labels. The proposed model learns pairs of interpretable latent representations for the nodes of directed graphs, and uses parameterized graph convolutional network (GCN) layers for its encoder and an asymmetric inner product decoder. Parameters in the encoder control the weighting of representations exchanged between neighboring nodes. We demonstrate the ability of the proposed model to learn meaningful latent embeddings and achieve superior performance on the directed link prediction task on several popular network datasets.
Cite
Text
Kollias et al. "Directed Graph Auto-Encoders." AAAI Conference on Artificial Intelligence, 2022. doi:10.1609/AAAI.V36I7.20682Markdown
[Kollias et al. "Directed Graph Auto-Encoders." AAAI Conference on Artificial Intelligence, 2022.](https://mlanthology.org/aaai/2022/kollias2022aaai-directed/) doi:10.1609/AAAI.V36I7.20682BibTeX
@inproceedings{kollias2022aaai-directed,
title = {{Directed Graph Auto-Encoders}},
author = {Kollias, Georgios and Kalantzis, Vasileios and Idé, Tsuyoshi and Lozano, Aurélie C. and Abe, Naoki},
booktitle = {AAAI Conference on Artificial Intelligence},
year = {2022},
pages = {7211-7219},
doi = {10.1609/AAAI.V36I7.20682},
url = {https://mlanthology.org/aaai/2022/kollias2022aaai-directed/}
}