AI Paper Atlas

Research map / Adversarial robustness and security / Backdoor and poisoning

Robust Contrastive Language-Image Pretraining against Data Poisoning and Backdoor Attacks

Wenhan Yang, Jingdong Gao, Baharan Mirzasoleiman

NeurIPS 2023 · Adversarial robustness and security › Backdoor and poisoning · approach: Backdoors in large multimodal models · 5 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 RoCLIP caption-pool matching instead of standard CLIP pre-trainingCLIP RN50 · Targeted image poisoningattack success rate: 93.75% -> 12.5%Improved
use RoCLIP instead of standard CLIP pre-trainingCLIP RN50 · HTBA backdoorattack success rate: 78% -> 0%Improved
RoCLIP vs CLIP on linear probe across ten datasetsCLIP RN50 · Food101, StanfordCars etc.linear-probe accuracy: 12.9 -> 20.35Improved
RoCLIP vs CLIP on zero-shotCLIP RN50 · 10 downstream datasetszero-shot accuracy: 34.9 -> 30.14No clear effect
lower matching frequency K from 3 to 2 for very strong poisoning (up to 1%)RoCLIP · CIFAR10 zero-shotzero-shot accuracy: 30.14 -> 21.3Hurt
lower K from 3 to 2 for stronger attacksRoCLIP · Targeted poisoning 0.5% poison rateattack success rate: 100% -> 0%Improved
add data augmentation to CLIP (no pool) (ablation)CLIP · 100K CC3M ablationpoison success rate: 75% -> 12.5%Improved
add data augmentation to CLIP (no pool) (ablation)CLIP · 100K CC3M ablationbackdoor success rate: 42.3% -> 0.33%Improved
add data augmentation to CLIP (ablation)CLIP · CIFAR10linear-probe accuracy: 62.1% -> 72.9%Improved
add data augmentation to CLIP (ablation)CLIP · CIFAR100linear-probe accuracy: 37.4% -> 48.4%Improved
increase caption pool size from 0.2% to 20% of dataset (ablation)RoCLIP · CIFAR10linear-probe accuracy: 64.5% -> 67.9%Improved
vary pool size 0.2%, 2%, 20% (ablation)RoCLIP · 100K ablationpoison success rate: 0% -> 0%No clear effect

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