What do larger image classifiers memorise?

TL;DR

This paper investigates how larger image classifiers memorize training data, revealing diverse memorization patterns across model sizes and showing that knowledge distillation reduces memorization and enhances generalization.

Contribution

It provides the first comprehensive empirical analysis of how model size affects memorization in image classification, challenging existing proxies for memorization measurement.

Findings

Most samples show decreased memorization with larger models.

Proxies for memorization scores fail to capture key trends.

Knowledge distillation inhibits memorization and improves generalization.

Abstract

The success of modern neural networks has prompted study of the connection between memorisation and generalisation: overparameterised models generalise well, despite being able to perfectly fit (memorise) completely random labels. To carefully study this issue, Feldman proposed a metric to quantify the degree of memorisation of individual training examples, and empirically computed the corresponding memorisation profile of a ResNet on image classification bench-marks. While an exciting first glimpse into what real-world models memorise, this leaves open a fundamental question: do larger neural models memorise more? We present a comprehensive empirical analysis of this question on image classification benchmarks. We find that training examples exhibit an unexpectedly diverse set of memorisation trajectories across model sizes: most samples experience decreased memorisation under larger models, while the rest exhibit cap-shaped or increasing memorisation. We show that various proxies for the Feldman memorization score fail to capture these fundamental trends. Lastly, we find that knowledge distillation, an effective and popular model compression technique, tends to inhibit memorisation, while also improving generalisation. Specifically, memorisation is mostly inhibited on examples with increasing memorisation trajectories, thus pointing at how distillation improves generalisation.