AI Paper Atlas

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

Human Uncertainty Makes Classification More Robust

Joshua C. Peterson, Ruairidh M. Battleday, Thomas L. Griffiths, Olga Russakovsky

ICCV 2019 · Adversarial robustness and security › Certified and empirical robustness · approach: Robustness-accuracy trade-off analysis · 233 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
train with human soft labels instead of one-hotCNN classifiers · CIFAR10accuracy: 83.5 -> 84.5Improved
train with human soft labels instead of one-hotCNN classifiers · ImageNet-Faraccuracy: 49.4 -> 51.4Improved
train with human soft labels instead of one-hotCNN classifiers · CIFAR10crossentropy: 0.7 -> 0.5Improved
train with human soft labels instead of one-hotCNN classifiers · ImageNet-Farcrossentropy: 2.9 -> 1.8Improved
train with human soft labels instead of one-hotCNN classifiers · CIFAR10H holdoutsecond-best accuracy: 5% better on averageImproved
use image-level human soft labels instead of class-level soft labels (ablation)ResNet fine-tuned on CIFAR10H · CIFAR10H holdoutcrossentropy: 0.42 -> 0.35Improved
use image-level human soft labels instead of mixup (ablation)PyramidNet fine-tuned · CIFAR10H holdoutcrossentropy: 0.49 -> 0.28Improved
fine-tune with human soft labels instead of ground-truth hard labels (ablation)Shake-Shake · CIFAR10H holdoutcrossentropy: 0.51 -> 0.26Improved
sample hard labels from human distribution instead of using soft targets (ablation)ResNet preact · CIFAR10.1v4crossentropy: 0.49 -> 0.42Improved
sample hard labels from human distribution instead of using soft targets (ablation)ResNet · CIFAR10H holdoutcrossentropy: 0.35 -> 0.35No clear effect
fine-tune with human soft labelsVGG under FGSM attack · CIFAR10 testcrossentropy: 7.9 -> 4.1Improved
fine-tune with human soft labelseight CNNs under FGSM attack · CIFAR10 testaccuracy: VGG 7% -> 8%, PyramidNet 22% -> 19%Mixed

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