AI Paper Atlas

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

Knowledge Distillation: Bad Models Can Be Good Role Models

Gal Kaplun, eran malach, Preetum Nakkiran, Shai Shalev-Shwartz

NeurIPS 2022 · Label-efficient and robust learning › Knowledge distillation · approach: Rethinking distillation objectives · 6 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
Replace one teacher with an ensemble of 10 teachers at inferenceResNet-18 teacher · CIFAR-10 testaccuracy: 81.3% -> 87.8%Improved
Distill using ensemble-labeled unlabeled examplesResNet-18 teacher ensemble · CIFAR-10 testaccuracy: 81.3% -> about 90%Improved
Distill using examples labeled by a randomly selected teacherResNet-18 teacher pool · CIFAR-10 testaccuracy: 81.3% -> about 90%Improved
Use ensemble labeling instead of randomly selected teacher labelingCIFAR-5m student distillation · CIFAR-10 testaccuracyNo clear effect
Increase the number of teachersEnsemble pseudo-labeling and random pseudo-labeling · CIFAR-10 testaccuracyImproved

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