Where Did Your Model Learn That? Label-free Influence for Self-supervised Learning

TL;DR

This paper introduces Influence-SSL, a label-free influence function for self-supervised learning that explains model predictions by analyzing data stability without labels, revealing insights into SSL models' data dependencies.

Contribution

The paper proposes Influence-SSL, a novel influence function tailored for SSL that operates without labels, bridging a gap in understanding data contributions in self-supervised models.

Findings

Influence-SSL effectively identifies influential training examples in SSL models.

SSL models respond differently to influential data compared to supervised models.

Applications include duplicate detection, outlier detection, and fairness analysis.

Abstract

Self-supervised learning (SSL) has revolutionized learning from large-scale unlabeled datasets, yet the intrinsic relationship between pretraining data and the learned representations remains poorly understood. Traditional supervised learning benefits from gradient-based data attribution tools like influence functions that measure the contribution of an individual data point to model predictions. However, existing definitions of influence rely on labels, making them unsuitable for SSL settings. We address this gap by introducing Influence-SSL, a novel and label-free approach for defining influence functions tailored to SSL. Our method harnesses the stability of learned representations against data augmentations to identify training examples that help explain model predictions. We provide both theoretical foundations and empirical evidence to show the utility of Influence-SSL in analyzing pre-trained SSL models. Our analysis reveals notable differences in how SSL models respond to influential data compared to supervised models. Finally, we validate the effectiveness of Influence-SSL through applications in duplicate detection, outlier identification and fairness analysis. Code is available at: \url{https://github.com/cryptonymous9/Influence-SSL}.