LEEP: A New Measure to Evaluate Transferability of Learned Representations

TL;DR

The paper introduces LEEP, a simple and efficient measure to evaluate the transferability of learned representations, capable of predicting transfer performance with high accuracy across various datasets and outperforming existing measures.

Contribution

It proposes LEEP, a novel transferability measure that is easy to compute and more accurate than previous methods, enhancing transfer learning evaluation.

Findings

LEEP can predict transfer performance and convergence speed.

Outperforms existing transferability measures like negative conditional entropy.

Achieves up to 30% improvement in correlation with actual transfer accuracy.

Abstract

We introduce a new measure to evaluate the transferability of representations learned by classifiers. Our measure, the Log Expected Empirical Prediction (LEEP), is simple and easy to compute: when given a classifier trained on a source data set, it only requires running the target data set through this classifier once. We analyze the properties of LEEP theoretically and demonstrate its effectiveness empirically. Our analysis shows that LEEP can predict the performance and convergence speed of both transfer and meta-transfer learning methods, even for small or imbalanced data. Moreover, LEEP outperforms recently proposed transferability measures such as negative conditional entropy and H scores. Notably, when transferring from ImageNet to CIFAR100, LEEP can achieve up to 30% improvement compared to the best competing method in terms of the correlations with actual transfer accuracy.