AI Paper Atlas

Research map / Adversarial robustness and security / Certified and empirical robustness

Generalization of Neural Combinatorial Solvers Through the Lens of Adversarial Robustness

Simon Geisler, Johanna Sommer, Jan Schuchardt, Aleksandar Bojchevski, Stephan Günnemann

ICLR 2022 · Adversarial robustness and security › Certified and empirical robustness · approach: Robust overfitting and regularization · 4 citations (OpenAlex)

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 changedWhereMeasuredEffect
Adversarially fine-tune NeuroSAT for 10 extra epochs using perturbed SAT instancesNeuroSAT · 10-40 Attackaccuracy: 50.0 ± 1.16 -> 54.0 ± 0.48Improved
Adversarially fine-tune NeuroSAT for 10 extra epochs using perturbed SAT instancesNeuroSAT · 50-100 Attackaccuracy: 39.4 ± 3.15 -> 44.4 ± 1.19Improved
Adversarially fine-tune NeuroSAT for 10 extra epochs using perturbed SAT instancesNeuroSAT · 100-300accuracy: 64.4 ± 1.53 -> 67.0 ± 0.74Improved
Adversarially fine-tune NeuroSAT for 10 extra epochs using perturbed SAT instancesNeuroSAT · SATLIBaccuracy: 66.1 ± 3.07 -> 63.7 ± 1.88Hurt
Train NeuroSAT on a 10% larger training set instead of regular trainingNeuroSAT · 10-40 Attackaccuracy: 50.0 ± 1.16 -> 49.6 ± 1.52No clear effect
Attack NeuroSAT with sound SAT-preserving literal perturbationsNeuroSAT · 10-40 satisfiable instancesaccuracy: 26%Hurt
Use a SAT perturbation budget of 0.2% of literalsNeuroSAT · SAT satisfiable instancesrecall: 0.2%Hurt
Add five sound adversarial nodes to DecisionTSP instancesDecisionTSP · TSP with n ∼ U(20,40) nodesaccuracyHurt
Add five sound adversarial nodes to ConvTSP instancesConvTSP · TSP with n = 20 nodesoptimality gap: 2%Hurt
Adversarially fine-tune NeuroSAT with a 5% budget on satisfiable instances (ablation)NeuroSAT · SAT satisfiable instancesaccuracy: 5%Hurt
Use adversarial fine-tuning rather than regular trainingNeuroSAT · 3-10accuracy: 92.3 ± 0.57 -> 93.0 ± 0.33Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy