Graph Robustness Benchmark
Graph Robustness Benchmark (GRB) provides scalable, general, unified, and reproducible evaluation on the adversarial robustness of graph machine learning, especially Graph Neural Networks (GNNs). GRB has elaborated datasets, unified evaluation pipeline, reproducible leaderboards, and modular coding framework, which facilitates a fair comparison among various attacks & defenses on GNNs and promotes future research in this field.
Install grb via pip:
pip install grb
Install grb via git:
git clone email@example.com:THUDM/grb.git cd grb pip install -e .
GRB provides all necessary components to ensure the reproducibility of evaluation results. Get datasets from link or download them by running the following script:
cd ./scripts sh download_dataset.sh
Get attack results (adversarial adjacency matrix and features) from link or download them by running the following script:
Get saved models (model weights) from link or download them by running the following script:
Usage of GRB Modules
Training a GNN model
An example of training Graph Convolutional Network (GCN) on grb-cora dataset.
import torch # pytorch backend from grb.dataset import Dataset from grb.model.torch import GCN from grb.utils.trainer import Trainer # Load data dataset = Dataset(name='grb-cora', mode='easy', feat_norm='arctan') # Build model model = GCN(in_features=dataset.num_features, out_features=dataset.num_classes, hidden_features=[64, 64]) # Training adam = torch.optim.Adam(model.parameters(), lr=0.01) trainer = Trainer(dataset=dataset, optimizer=adam, loss=torch.nn.functional.nll_loss) trainer.train(model=model, n_epoch=200, dropout=0.5, train_mode='inductive')
An example of applying Topological Defective Graph Injection Attack (TDGIA) on trained GCN model.
from grb.attack.tdgia import TDGIA # Attack configuration tdgia = TDGIA(lr=0.01, n_epoch=10, n_inject_max=20, n_edge_max=20, feat_lim_min=-0.9, feat_lim_max=0.9, sequential_step=0.2) # Apply attack rst = tdgia.attack(model=model, adj=dataset.adj, features=dataset.features, target_mask=dataset.test_mask) # Get modified adj and features adj_attack, features_attack = rst
GRB provides a unified evaluation scenario for fair comparisons between attacks and defenses. The scenario is Black-box, Evasion, Inductive, Injection. Take the case of a citation-graph classification system for example. The platform collects labeled data from previous papers and trains a GNN model. When a batch of new papers are submitted, it updates the graph and uses the trained model to predict labels for them.
- Black-box: Both the attacker and the defender have no knowledge about the applied methods each other uses.
- Evasion: GNNs are already trained in trusted data (e.g. authenticated users), which are untouched by the attackers but might have natural noises. Thus, attacks will only happen during the inference phase.
- Inductive: GNNs are used to classify unseen data (e.g. new users), i.e. validation or test data are unseen during training, which requires GNNs to generalize to out of distribution data.
- Injection: The attackers can only inject new nodes but not modify the target nodes directly. Since it is usually hard to hack into users' accounts and modify their profiles. However, it is easier to create fake accounts and connect them to existing users.
GRB maintains leaderboards that permits a fair comparision across various attacks and defenses. To ensure the reproducibility, we provide all necessary information including datasets, attack results, saved models, etc. Besides, all results on the leaderboards can be easily reproduced by running the following scripts (e.g. leaderboard for grb-cora dataset):
sh run_leaderboard_pipeline.sh -d grb-cora -g 0 -s ./leaderboard -n 0 Usage: run_leaderboard_pipeline.sh [-d <string>] [-g <int>] [-s <string>] [-n <int>] Pipeline for reproducing leaderboard on the chosen dataset. -h Display help message. -d Choose a dataset. -s Set a directory to save leaderboard files. -n Choose the number of an attack from 0 to 9. -g Choose a GPU device. -1 for CPU.
We welcome researchers to submit new methods including attacks, defenses, or new GNNS to enrich the GRB leaderboard. For future submissions, one should follow the GRB Evaluation Rules and respect the reproducibility.
Please submit your methods via the google form GRB submission. Our team will verify the result within a week.
In case of any problem, please contact us via email: firstname.lastname@example.org or email@example.com. We also welcome researchers to join our Google Group for further discussion on the adversarial robustness of graph machine learning.
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size grb-0.1.0.tar.gz (40.3 kB)||File type Source||Python version None||Upload date||Hashes View|