This code reproduces the result of [1] in PyTorch, and the code derives from https://github.com/Diego999/pyGAT, which is a PyTorch implementation of [2]. The classification accuracy is 95.14%.

Run GAT_EFA:

sh ./script/GAT_EFA.sh

You may also run other methods (MLP, GCN, GAT) with .sh files in path ./script.

[1] Yun X, Zhang Y, Ye J, Liu C. Online Handwritten Diagram Recognition with Graph Attention Networks[C]. //Chao Y, Barnes N, Chen Baoquan, et al. Image and Graphics – 10th International Conference, ICIG 2019, Beijing, China, August 23-25, 2019, Proceedings, Part I. Beijing, China: Springer, 2019: 232-244. [2] Veličković P, Cucurull G, Casanova A, et al. Graph Attention Networks[J]. Robust Photometric Stereo Using Learned Image and Gradient Dictionaries: 2017, abs/1710.00002: 1710.10903.