AI Paper Atlas

Research map / Learning theory and algorithms / Combinatorial and search optimization

Tight Neural Network Verification via Semidefinite Relaxations and Linear Reformulations

Jianglin Lan, Yang Zheng, Alessio Lomuscio

AAAI 2022 · Learning theory and algorithms › Combinatorial and search optimization · approach: Semidefinite relaxations for verification · 13 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
Add RLT-generated linear cuts to LayerSDPRLT-SDP · MNIST, MLP-Advcertified robustness: 80 | 91 -> 88 | 94Improved
Add RLT-generated linear cuts to LayerSDPRLT-SDP · MNIST, MLP-LPcertified robustness: 80 | 80 -> 80 | 80No clear effect
Add RLT-generated linear cuts to LayerSDPRLT-SDP · MNIST, MLP-SDPcertified robustness: 80 | 84 -> 84 | 84Improved
Add RLT-generated linear cuts to LayerSDPRLT-SDP · MNIST, 6 × 100certified robustness: 60 | 75 -> 82 | 90Improved
Add RLT-generated linear cuts to LayerSDPRLT-SDP · MNIST, 9 × 100certified robustness: 22 | 35 -> 56 | 70Improved
Use RLT-SDP instead of IBPRLT-SDP · MNIST, 8×1024-0.1certified robustness: 67 -> 82Improved
Use RLT-SDP instead of IBPRLT-SDP · MNIST, 8×1024-0.3certified robustness: 7 -> 26Improved
Add a larger percentage of RLT linear cuts (ablation)RLT-SDP · random ReLU networks, feasible output regionover-approximation tightnessImproved
Add more RLT linear cuts (ablation)RLT-SDP · random ReLU networks, runtimeruntime increaseHurt
Use the first 20% of RLT linear cuts (ablation)RLT-SDP · random ReLU networks, feasible output regionover-approximation tightnessImproved

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