WeShap: Weak Supervision Source Evaluation with Shapley Values

TL;DR

WeShap introduces a Shapley value-based metric for evaluating and improving weak supervision sources, enabling better understanding and refinement of programmatic labeling pipelines to enhance machine learning model accuracy.

Contribution

This paper proposes WeShap, a novel, efficient Shapley value-based evaluation metric for weak supervision sources, with demonstrated versatility and generalizability across different PWS pipelines.

Findings

WeShap effectively identifies beneficial and detrimental labeling functions.

Refining PWS pipelines with WeShap improves downstream model accuracy by an average of 5.0 points.

WeShap generalizes well to various PWS pipeline configurations.

Abstract

Efficient data annotation stands as a significant bottleneck in training contemporary machine learning models. The Programmatic Weak Supervision (PWS) pipeline presents a solution by utilizing multiple weak supervision sources to automatically label data, thereby expediting the annotation process. Given the varied contributions of these weak supervision sources to the accuracy of PWS, it is imperative to employ a robust and efficient metric for their evaluation. This is crucial not only for understanding the behavior and performance of the PWS pipeline but also for facilitating corrective measures. In our study, we introduce WeShap values as an evaluation metric, which quantifies the average contribution of weak supervision sources within a proxy PWS pipeline, leveraging the theoretical underpinnings of Shapley values. We demonstrate efficient computation of WeShap values using dynamic programming, achieving quadratic computational complexity relative to the number of weak supervision sources. Our experiments demonstrate the versatility of WeShap values across various applications, including the identification of beneficial or detrimental labeling functions, refinement of the PWS pipeline, and rectification of mislabeled data. Furthermore, WeShap values aid in comprehending the behavior of the PWS pipeline and scrutinizing specific instances of mislabeled data. Although initially derived from a specific proxy PWS pipeline, we empirically demonstrate the generalizability of WeShap values to other PWS pipeline configurations. Our findings indicate a noteworthy average improvement of 5.0 points in downstream model accuracy through the revision of the PWS pipeline compared to previous state-of-the-art methods, underscoring the efficacy of WeShap values in enhancing data quality for training machine learning models.