https://github.com/virresh/viking
Adversarial Attack on Network Embeddings via Supervised Network Poisoning
https://github.com/virresh/viking
Last synced: 11 months ago
JSON representation
Adversarial Attack on Network Embeddings via Supervised Network Poisoning
- Host: GitHub
- URL: https://github.com/virresh/viking
- Owner: virresh
- Created: 2021-02-14T14:39:34.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2021-02-14T14:41:25.000Z (over 5 years ago)
- Last Synced: 2025-03-23T19:39:07.382Z (about 1 year ago)
- Language: Python
- Size: 3.49 MB
- Stars: 5
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# VIKING
Code for paper accepted at PAKDD 2021,
VIKING: Adversarial Attack on Network Embeddings via Supervised Network Poisoning
Authors: Viresh Gupta, Tanmoy Chakraborty
This repository has two main components:
1. code - This folder contains the code for generating VIKING and VIKING^s perturbations. Additionally it also contains sample implementations of baselines and evaluation methods for node classification and link prediction tasks.
2. data - This folder contains the three datasets used for reporting the given results.
### Requirements for running the code
Make sure you have installed node2vec and LINE binaries and they are avalaible on path.
(i.e `node2vec` and `line` should not print error on terminal).
If you face errors while running line, make sure you have all required c-libraries on path.
Also ensure `pytorch` and `dgl` library is installed.
### Running instructions
From the root directory of this repository, call main.py as follows:
```bash
python code/simple_all.py cora lp
```
Since the results are quite long, it's ideal to save output to a file and later on tail the file for required output.
### Acknowledgements
Some code has been borrowed from previous research work available at https://github.com/abojchevski/node_embedding_attack
The same has been marked in the code files where applicable.