Domino: Discovering Systematic Errors with Cross-Modal Embeddings

TL;DR

This paper introduces Domino, a novel cross-modal embedding-based slice discovery method that accurately identifies systematic errors in high-dimensional data and provides natural language descriptions, outperforming prior approaches.

Contribution

We develop a quantitative evaluation framework for slice discovery methods and propose Domino, which leverages cross-modal embeddings and a new error-aware model for coherent slice identification and description.

Findings

Domino identifies 36% of slices, a 12% improvement over prior methods.

It correctly generates slice names in 35% of cases.

The evaluation framework covers 1,235 settings across multiple data domains.

Abstract

Machine learning models that achieve high overall accuracy often make systematic errors on important subsets (or slices) of data. Identifying underperforming slices is particularly challenging when working with high-dimensional inputs (e.g. images, audio), where important slices are often unlabeled. In order to address this issue, recent studies have proposed automated slice discovery methods (SDMs), which leverage learned model representations to mine input data for slices on which a model performs poorly. To be useful to a practitioner, these methods must identify slices that are both underperforming and coherent (i.e. united by a human-understandable concept). However, no quantitative evaluation framework currently exists for rigorously assessing SDMs with respect to these criteria. Additionally, prior qualitative evaluations have shown that SDMs often identify slices that are incoherent. In this work, we address these challenges by first designing a principled evaluation framework that enables a quantitative comparison of SDMs across 1,235 slice discovery settings in three input domains (natural images, medical images, and time-series data). Then, motivated by the recent development of powerful cross-modal representation learning approaches, we present Domino, an SDM that leverages cross-modal embeddings and a novel error-aware mixture model to discover and describe coherent slices. We find that Domino accurately identifies 36% of the 1,235 slices in our framework - a 12 percentage point improvement over prior methods. Further, Domino is the first SDM that can provide natural language descriptions of identified slices, correctly generating the exact name of the slice in 35% of settings.