Description

- t-SNE for graph nodes neighbors

• t-SNE learns a two-dimensional embedding vector that preserves a certain neighboring structure of data represented as a high-dimensional vector, and expresses high-dimensional data as a two-dimensional map.
• We add two py files (get_neighbor_index.py, select_feature.py) to get neighbors index list of target node and select embedding vector appropriated for index list.
• Plot the x-axis and y-axis of t-SNE classified labels

- Examples

Contents

- get_neighbor_index.py

• Save neighbors index list txt of target node using NetworkX

- select_feature.py

• Select embedding vector apropriated to index list
• index_text mode: read index list txt, index_range mode: set range of index list

- tsne_torch.py

• Read feature vectors & labels and learning t-SNE
• Plot axies of t-SNE
• Using CUDA tensor

- Order of execution: get_neighbor_index.py -> select_feature.py -> tsne_torch.py

How to use

python get_neighbor_index.py --target_node u27192 --hop 1

python select_feature.py --index_text 1

python tsne_torch.py --max_iter 1000

Datasets

- Amazon-Book Dataset

https://jmcauley.ucsd.edu/data/amazon/

References

- NetworkX Library

https://networkx.github.io/documentation/stable/index.html

- t-SNE Pytorch

https://github.com/mxl1990/tsne-pytorch

- sklearn t-SNE

https://github.com/qbxlvnf11/machine-learning-basic/blob/master/Character-level_Word-level_t-SNE.ipynb

Author

- LinkedIn: https://www.linkedin.com/in/taeyong-kong-016bb2154

- Blog URL: https://blog.naver.com/qbxlvnf11

- Email: qbxlvnf11@google.com, qbxlvnf11@naver.com