AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

LEEP: A New Measure to Evaluate Transferability of Learned Representations

Cuong Nguyen, Tal Hassner, Matthias Seeger, Cedric Archambeau

ICML 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 69 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
Use LEEP instead of NCE to estimate transfer performanceRe-train head, CIFAR100 large balanced from ImageNet · CIFAR100, large balancedPearson correlation with test accuracy: 0.974 vs 0.973No clear effect
Use LEEP instead of NCE to estimate transfer performanceRe-train head, CIFAR100 small balanced from ImageNet · CIFAR100, small balancedPearson correlation with test accuracy: 0.715 -> 0.798Improved
Use LEEP instead of NCE to estimate transfer performanceRe-train head, CIFAR100 small balanced noisy from ImageNet · CIFAR100, small balanced noisyPearson correlation with test accuracy: 0.579 -> 0.612Improved
Use LEEP instead of NCE to estimate transfer performanceRe-train head, CIFAR100 small imbalanced from ImageNet · CIFAR100, small imbalancedPearson correlation with test F1: 0.484 -> 0.522Improved
Use LEEP instead of H score to estimate transfer performanceRe-train head, CIFAR100 small balanced from ImageNet · CIFAR100, small balancedPearson correlation with test accuracy: 0.026* -> 0.798Improved
Use LEEP instead of H score to estimate transfer performanceFine-tune, CIFAR100 small balanced from ImageNet · CIFAR100, small balancedPearson correlation with test accuracy: -0.029* -> 0.762Improved
Use LEEP instead of NCE to estimate meta-transfer performanceCNAPS, CIFAR100 small balanced from ImageNet · CIFAR100, small balancedPearson correlation with test accuracy: 0.310 -> 0.591Improved
Use LEEP instead of NCE to estimate transfer performanceFine-tune, CIFAR100 small balanced from CIFAR10 · CIFAR100, small balancedPearson correlation with test accuracy: 0.401 -> 0.396Hurt
Use LEEP instead of NCE to estimate transfer performanceFine-tune, CIFAR100 large balanced from ImageNet · CIFAR100, large balancedPearson correlation with test accuracy: 0.945 -> 0.944Hurt
Use LEEP instead of H score to estimate transfer performanceRe-train head, FashionMNIST small balanced from CIFAR10 · FashionMNIST, small balancedPearson correlation with test accuracy: 0.787 -> 0.518Hurt
Use LEEP instead of H score to estimate transfer performanceFine-tune, FashionMNIST small balanced from CIFAR10 · FashionMNIST, small balancedPearson correlation with test accuracy: 0.826 -> 0.339Hurt
Use LEEP instead of H score to estimate transfer performanceFine-tune, CIFAR100 small imbalanced from CIFAR10 · CIFAR100, small imbalancedPearson correlation with test F1: 0.758 -> 0.597Hurt

Later papers that compared with this method

12 later accepted papers report LEEP as a baseline in their results tables; 11 of them report a better number on most of the results they share with it.

4 more not listed.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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