AI Paper Atlas

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

Part-Based Models Improve Adversarial Robustness

Chawin Sitawarin, Kornrapat Pongmala, Yizheng Chen, Nicholas Carlini, David Wagner

ICLR 2023 · Adversarial robustness and security › Certified and empirical robustness · approach: Robustness-accuracy trade-off analysis · 2 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
Use part-level segmentation supervision with the downsampled part model instead of omitting segmentation labels (ablation)Downsampled Part Model · Part-ImageNetclean accuracy: 76.9 -> 85.6Improved
Use part-level segmentation supervision with the downsampled part model instead of omitting segmentation labels (ablation)Downsampled Part Model · Part-ImageNetadversarial accuracy: 39.6 -> 39.4No clear effect
Use part-level segmentation supervision with the bounding-box part model instead of omitting segmentation labels (ablation)Bounding-Box Part Model · Part-ImageNetclean accuracy: 78.1 -> 86.5Improved
Use part-level segmentation supervision with the bounding-box part model instead of omitting segmentation labels (ablation)Bounding-Box Part Model · Part-ImageNetadversarial accuracy: 39.9 -> 39.2No clear effect
Replace segmentation masks with part bounding-box labels (ablation)Part-based models · Part-ImageNetclean accuracy: 85.6 -> 84.1Hurt
Replace segmentation masks with part centroid labels (ablation)Part-based models · Part-ImageNetclean accuracy: 85.6 -> 82.6Hurt
Replace segmentation masks with part bounding-box labels (ablation)Part-based models · Cityscapesclean accuracy: 94.8 -> 95.4Mixed
Replace segmentation masks with part bounding-box labels (ablation)Part-based models · PASCAL-Partclean accuracy: 77.3 -> 66.2Hurt
Use part segmentation supervision instead of object-level segmentation labels (ablation)Part-based model · Part-ImageNetaccuracyImproved
Reduce the fraction of samples with part segmentation annotations (ablation)Part-based models · Part-ImageNetadversarial accuracyNo clear effect
Use pseudo-labels from another segmentation model to reduce human segmentation labels (ablation)Part-based models · Part-ImageNetaccuracy and robustnessNo clear effect
Increase segmentation-loss weight (ablation)Part-based models · Part-ImageNetadversarial accuracyMixed

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