The Graph Autoencoder (GAE) model takes two matrices: an adjacency matrix $$A$$ and a feature matrix $$X$$. The feature matrix $$X$$ has size $$n \times d$$, where $$n$$ is the number of vertices/concepts and $$d$$ is the number of input features per concept.

University of Michigan - Ann Arbor

Google

The author used the non-probabilistic Graph autoencoder(GAE) and the Variational Graph Autoencoder(VGAE) with 200 epochs, a learning rate of 0.01, two-layers(32 hidden units in the first and 16 in the second). (Kipf and Welling 2016)

What Should I Learn First: Introducing LectureBank for NLP Education and Prerequisite Chain Learning: Prerequisite chain learning GCN models

Li, I., Fabbri, A. R., Tung, R. R., & Radev, D. R. (2019, July). What should i learn first: Introducing lecturebank for nlp education and prerequisite chain learning. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 33, pp. 6674-6681).

Paper: https://www.aaai.org/ojs/index.php/AAAI/article/view/4638

Github: https://github.com/Yale-LILY/LectureBank

What Should I Learn First: Introducing LectureBank for NLP Education and Prerequisite Chain Learning: Reference

Kipf, T. N., and Welling, M. 2016b. Variational graph autoencoders. NIPS Workshop on Bayesian Deep Learning.

https://arxiv.org/pdf/1611.07308.pdf

What Should I Learn First: Introducing LectureBank for NLP Education and Prerequisite Chain Learning: Reference for GCNs

VGAE is an extension to the GAE model by taking latent variables.

Learn Before

Related