Features • Installation • Examples
Pytorch Geometric extension library with additional graph sampling algorithms.
Supports:
- Node2Vec 1 (
random_walk) - Temporal Random Walk (
temporal_random_walk) - Biased Temporal Random Walk (CTDNE) 2 (
biased_tempo_random_walk) - Negative Sampling (
negative_sample_neighbors_homogenousandnegative_sample_neighbors_heterogenous) - GraphSAGE + budget sampling (
budget_sampling) - Temporal Heterogenous Graph Transformer (HGT) sampling [^4] (
hgt_sampling) - GraphSAGE 3 (
neighbor_sampling_heterogenousandneighbor_sampling_homogenous)
If you are using CPU only installation of pytoch, install tch-geometric using pip:
pip install tch_geometric
If you are using GPU accelerated pytoch you need to use conda
conda install egordm::tch_geometric
You can specify the cuda version explicity as follows (python 3.9, pytorch 1.11, cudnn 113):
conda install egordm::tch_geometric=0.1.0=py39_torch_1.11.0_cu113
THe examples can be found in the examples folder
Footnotes
-
A. Grover and J. Leskovec, “node2vec: Scalable Feature Learning for Networks,” in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, New York, NY, USA, Aug. 2016, pp. 855–864. doi: 10.1145/2939672.2939754. ↩
-
G. H. Nguyen, J. B. Lee, R. A. Rossi, N. K. Ahmed, E. Koh, and S. Kim, “Continuous-Time Dynamic Network Embeddings,” in Companion of the The Web Conference 2018 on The Web Conference 2018 - WWW ’18, Lyon, France, 2018, pp. 969–976. doi: 10.1145/3184558.3191526. ↩
-
W. L. Hamilton, R. Ying, and J. Leskovec, “Inductive representation learning on large graphs,” in Proceedings of the 31st International Conference on Neural Information Processing Systems, Red Hook, NY, USA, Dec. 2017, pp. 1025–1035. [^4] Z. Hu, Y. Dong, K. Wang, and Y. Sun, “Heterogeneous Graph Transformer (tempo),” arXiv:2003.01332 [cs, stat], Mar. 2020, Accessed: Mar. 09, 2022. [Online]. Available: http://arxiv.org/abs/2003.01332 ↩