AI Paper Atlas

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

On the Decision Boundaries of Neural Networks: A Tropical Geometry Perspective

Motasem Alfarra, Adel Bibi, Hasan Abed Al Kader Hammoud, Mohamed Gaafar, Bernard Ghanem

TPAMI 2022 · Adversarial robustness and security › Certified and empirical robustness · approach: Decision-boundary geometry of robustness · 46 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
prune with tropical decision-boundary objective instead of Class Blind/Uniform/DistributionAlexNet classifier · SVHNpruning-accuracy AUC: better in AUC by 3%Improved
prune with tropical objectiveAlexNet classifier · CIFAR10pruning-accuracy AUC: better in AUC by 3%Improved
prune with tropical objectiveAlexNet classifier · CIFAR100pruning-accuracy AUC: better in AUC by 2%Improved
prune with tropical objectiveVGG16 classifier · CIFAR100pruning-accuracy AUCHurt
prune with tropical objectiveVGG16 classifier · SVHN and CIFAR10pruning-accuracy AUCNo clear effect
tropical pruning instead of SM heuristicLeNet · MNIST, 90% neurons kepttest accuracy: 95.71 -> 98.82Improved
tropical pruning instead of SM heuristicLeNet · MNIST, 25% neurons kepttest accuracy: 91.04 -> 98.36Improved
tropical pruning instead of SM heuristicLeNet · MNIST, 5% neurons kepttest accuracy: 92.93 -> 94.91Improved
add decision-boundary polytope perturbation (eps2=1.0) to structured attack (ablation)structured adversarial attack · MNIST, eps1=0.1robust accuracy: 44.8 -> 46.1Hurt
add decision-boundary polytope perturbation (eps2=2.0) to structured attack (ablation)structured adversarial attack · MNIST, eps1=0.1robust accuracy: 44.8 -> 43.8Improved
add decision-boundary polytope perturbation (eps2=2.0) to structured attack (ablation)structured adversarial attack · MNIST, eps1=0.2robust accuracy: 12.6 -> 11.0Improved
initialize pruned network with the original (lottery ticket) initialization instead of Xavier/Gaussian (ablation)pruned 2-layer network · synthetic 2D binary dataorientation of decision boundaries polytopeImproved

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