Research map / Adversarial robustness and security / Black-box and transfer attacks
Adversarial Attacks and Robust Training for Hypergraph Neural Networks
Naheed Anjum Arafat, Debabrota Basu, Yulia Gel, Danda B. Rawat
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use MeLA-PGD multi-step attack instead of single-step MeLA-FGSM | MeLA attack on AllSetTransformer · Citeseer poisoning | accuracy: 61.59 -> 28.82 | Improved |
| use MeLA-PGD instead of MeLA-FGSM | MeLA attack on HGNN · Cora-CA poisoning | accuracy: 73.44 -> 72.64 | Improved |
| use MeLA-FGSM instead of RandFlip | attack on HGNN · Citeseer poisoning | accuracy drop %: 13.2 -> 7.2 | Mixed |
| use MeLA-PGD instead of strongest baseline attacks | attack on HyperMLP · Citeseer poisoning | accuracy drop %: 4.0 -> 15.4 | Improved |
| use MeLA-PGD in evasion instead of random baselines | attack on AllSetTransformer · Cora-CA evasion | accuracy drop %: 4.2 -> 73.5 | Improved |
| train with MeLA-D adversarial training | HyperMLP under MeLA-PGD poisoning · Citeseer | accuracy: 41.69 -> 69.81 | Improved |
| train with MeLA-D adversarial training | HyperMLP under MeLA-PGD poisoning · Cora-CA | accuracy: 25.05 -> 66.65 | Improved |
| train with MeLA-D adversarial training | HyperMLP under MeLA-FGSM poisoning · Cora | accuracy: 39.05 -> 68.33 | Improved |
| train with MeLA-D adversarial training | HyperMLP under RandFlip poisoning · Citeseer | accuracy: 70.48 -> 72.80 | Improved |
| train with MeLA-D adversarial training | HyperMLP under RandFeat poisoning · Cora-CA | accuracy: 69.69 -> 73.86 | Improved |
| apply mini-batch MeLA-PGD to large hypergraphs (evasion) | HGNN · Flickr | accuracy: 89.53 -> 35.75 | Hurt |
| apply mini-batch MeLA-PGD to large hypergraphs (evasion) | HGNN · Walmart | accuracy: 97.70 -> 95.96 | No clear effect |
Closest papers by meaning
- Hypergraph Attacks via Injecting Homogeneous Nodes into Elite Hyperedges (AAAI 2025 · similarity 0.89)
- HyperNear: Unnoticeable Node Injection Attacks on Hypergraph Neural Networks (ICML 2025 · similarity 0.89)
- Bounding the Expected Robustness of Graph Neural Networks Subject to Node Feature Attacks (ICLR 2024 · similarity 0.88)
- Value at Adversarial Risk: A Graph Defense Strategy against Cost-Aware Attacks (AAAI 2024 · similarity 0.88)
- A Simple and Yet Fairly Effective Defense for Graph Neural Networks (AAAI 2024 · similarity 0.87)
- If You Want to Be Robust, Be Wary of Initialization (NeurIPS 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
