AI Paper Atlas

Research map / Label-efficient and robust learning / Noisy labels

Symmetric Cross Entropy for Robust Learning With Noisy Labels

Yisen Wang, Xingjun Ma, Zaiyi Chen, Yuan Luo, Jinfeng Yi, James Bailey

ICCV 2019 · Label-efficient and robust learning › Noisy labels · approach: Sample selection and co-training · 1,088 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 reverse cross entropy (SL) to CECE, 8-layer CNN · CIFAR-10 (60% symmetric noise)test accuracy (%): 66.62 -> 80.07Improved
add reverse cross entropy (SL) to CECE, 8-layer CNN · CIFAR-10 (40% symmetric noise)test accuracy (%): 78.70 -> 85.34Improved
add reverse cross entropy (SL) to CECE, ResNet-44 · CIFAR-100 (60% symmetric noise)test accuracy (%): 25.39 -> 41.47Improved
add reverse cross entropy (SL) to CECE, 4-layer CNN · MNIST (60% symmetric noise)test accuracy (%): 46.54 -> 97.40Improved
add reverse cross entropy (SL) to CECE, ResNet-50 · Clothing1Maccuracy (%): 68.80 -> 71.02Improved
add SL to CE to improve learning on hard classes (ablation)CE, 8-layer CNN · CIFAR-10 (40% symmetric noise), class 3 clean portionprediction confidence: 0.5 -> 0.95Improved
add RCE term to ForwardForward, 8-layer CNN · CIFAR-10 (40% symmetric noise)test accuracy (%): 81.37 -> 84.54Improved
add RCE term to ForwardForward, 8-layer CNN · CIFAR-10 (60% symmetric noise)test accuracy (%): 73.59 -> 79.64Improved
add RCE term to ForwardForward, 8-layer CNN · CIFAR-10 (40% asymmetric noise)test accuracy (%): 85.73 -> 86.22Improved
replace LSR with SL lossLSR, 8-layer CNN · CIFAR-10 (40% symmetric noise)test accuracy (%): 78.41 -> 85.20Improved
replace LSR with SL lossLSR, 8-layer CNN · CIFAR-10 (40% asymmetric noise)test accuracy (%): 77.88 -> 80.99Improved
upscale the CE term (2CE, 5CE) instead of adding RCE (ablation)CE, 8-layer CNN · CIFAR-10 (60% symmetric noise)test accuracyHurt

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