AI Paper Atlas

Research map / Label-efficient and robust learning / Knowledge distillation

Be Your Own Teacher: Improve the Performance of Convolutional Neural Networks via Self Distillation

Linfeng Zhang, Jiebo Song, Anni Gao, Jingwei Chen, Chenglong Bao, Kaisheng Ma

ICCV 2019 · Label-efficient and robust learning › Knowledge distillation · approach: Rethinking distillation objectives · 1,082 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 self distillation (auxiliary classifiers trained from deepest one)ResNet50 · CIFAR100top-1 accuracy (%): 77.68 -> 80.56Improved
add self distillationResNet50 · ImageNet valtop-1 accuracy (%): 73.56 -> 75.24Improved
add self distillationResNet18 · ImageNet valtop-1 accuracy (%): 68.12 -> 69.84Improved
add self distillationResNeXt29-8 · CIFAR100top-1 accuracy (%): 81.29 -> 81.51Improved
use self distillation instead of deep supervision (DSN) (ablation)ResNet50 · CIFAR100top-1 accuracy (%): 80.27 -> 80.56Improved
use self distillation instead of deep supervision (DSN), shallowest classifier (ablation)ResNet152 · CIFAR100classifier 1/4 accuracy (%): 67.60 -> 68.84Improved
use self distillation instead of traditional KD with teacher ResNet152 (ablation)ResNet50 student · CIFAR100top-1 accuracy (%): 79.33 -> 80.56Improved
use self distillation instead of AT with a WideResNet44-8 teacher (ablation)WideResNet20-8 · CIFAR100top-1 accuracy (%): 80.65 -> 80.92Improved
replace pre-trained teacher training with one-step self distillation (ablation)ResNet50 · CIFAR100training time: 26.98 hours -> 5.87 hoursImproved
add ensemble of all classifiers' softmax outputs (ablation)self distillation ResNet50 · ImageNet valtop-1 accuracy (%): 75.24 -> 74.73Hurt
add ensemble of all classifiers' softmax outputs (ablation)self distillation ResNet50 · CIFAR100top-1 accuracy (%): 80.56 -> 81.04Improved
exit at shallow classifier 1/4 instead of the full network (ablation)self distillation ResNet50 · CIFAR100top-1 accuracy (%): 77.68 -> 68.23Hurt

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