AI Paper Atlas

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

DOT: A Distillation-Oriented Trainer

Borui Zhao, Quan Cui, Renjie Song, Jiajun Liang

ICCV 2023 · Label-efficient and robust learning › Knowledge distillation · approach: Rethinking distillation objectives · 17 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
apply larger momentum to distillation gradient than task gradient (DOT)KD · CIFAR-100 (ResNet32x4 -> ResNet8x4)top-1 accuracy (%): 73.33 -> 75.12Improved
apply DOT to KD on a harder datasetKD · Tiny-ImageNet (ResNet18 -> MobileNetV2)top-1 accuracy (%): 58.35 -> 64.01Improved
apply DOT to KD on ImageNet-1kKD · ImageNet-1k (ResNet50 -> MobileNetV1)top-1 accuracy (%): 70.50 -> 73.09Improved
apply DOT to DKDDKD · ImageNet-1k (ResNet50 -> MobileNetV1)top-1 accuracy (%): 72.05 -> 73.33Improved
apply DOT to feature-based CRDCRD · CIFAR-100 (ResNet32x4 -> ResNet8x4)top-1 accuracy (%): 75.51 -> 75.99Improved
use negative Delta so task loss has larger momentum (ablation)DOT · CIFAR-100 (ResNet32x4 -> ResNet8x4)top-1 accuracy (%): 73.33 -> 61.58Hurt
use positive Delta between 0.025 and 0.09 (ablation)KD · CIFAR-100 (ResNet32x4 -> ResNet8x4)top-1 accuracy (%): 73.33 -> 75.12Improved
tune the SGD momentum mu without DOT (ablation)KD · CIFAR-100 (ResNet32x4 -> ResNet8x4)top-1 accuracy (%): 73.33 -> 73.66No clear effect
adjust the task loss weight alpha (ablation)KD · CIFAR-100 (ResNet32x4 -> ResNet8x4)top-1 accuracy (%): 73.33 -> 73.56No clear effect
adjust the task loss weight alpha (ablation)KD · Tiny-ImageNet (ResNet18 -> MobileNetV2)top-1 accuracy (%): 58.35 -> 59.23No clear effect
train KD for longer epochs (240 to 300) (ablation)KD · CIFAR-100 (ResNet32x4 -> ResNet8x4)training task loss: 0.3844 -> 0.3818No clear effect
apply DOT to DKD on CIFAR-100 ShuffleNetV2 student, top-5DKD · CIFAR-100 (ResNet32x4 -> ShuffleNetV2)top-5 accuracy (%): (-0.06)Hurt

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