GSM: Inductive Learning on Dynamic Graph Embeddings

P. 85–99.

Ананьева М. Е., Макаров И. А., Pendiukhov M.

In this paper, we study the problem of learning graph embeddings for dynamic networks and the ability to generalize to unseen nodes called inductive learning. Firstly, we overview the state-of-the-art methods and techniques for constructing graph embeddings and learning algorithms for both transductive and inductive approaches. Secondly, we propose an improved GSM based on GraphSAGE algorithm and set up the experiments on datasets CORA, Reddit, and HSEcite, which is collected from Scopus citation database across the authors with affiliation to NRU HSE in 2011–2017. The results show that our three-layer model with attention-based aggregation function, added normalization layers, regularization (dropout) outperforms suggested by the respective authors’ GraphSAGE models with mean, LSTM, and pool aggregation functions, thus giving more insight into possible ways to improve inducting learning model based on GraphSAGE model.

Ключевые слова: Graph Embeddings, графовые эмбеддинги, Dynamic graphs, Inductive learning approach

В книге

Network Algorithms, Data Mining, and Applications. Springer Proceedings in Mathematics & Statistics

Springer, 2020.