Snorkel: rapid training data creation with weak supervision

TLDR

Labeling training data is increasingly the largest bottleneck in deploying machine learning systems. The paper introduces Snorkel, a system that lets users train state‑of‑the‑art models without hand‑labeling data, and explores trade‑offs by proposing an optimizer for faster pipeline execution. Snorkel uses user‑written labeling functions to express heuristics, denoises their outputs via data programming without ground truth, and provides a flexible interface built from industry collaboration. User studies show experts build models 2.8× faster and improve performance by 45.5% compared to hand labeling, and in collaborations with the VA and FDA, Snorkel boosts predictive performance by 132% over heuristics and matches large hand‑curated sets within 3.6%.

Abstract

Labeling training data is increasingly the largest bottleneck in deploying machine learning systems. We present Snorkel, a first-of-its-kind system that enables users to train state-of-the-art models without hand labeling any training data. Instead, users write labeling functions that express arbitrary heuristics, which can have unknown accuracies and correlations. Snorkel denoises their outputs without access to ground truth by incorporating the first end-to-end implementation of our recently proposed machine learning paradigm, data programming. We present a flexible interface layer for writing labeling functions based on our experience over the past year collaborating with companies, agencies, and research laboratories. In a user study, subject matter experts build models 2.8× faster and increase predictive performance an average 45.5% versus seven hours of hand labeling. We study the modeling trade-offs in this new setting and propose an optimizer for automating trade-off decisions that gives up to 1.8× speedup per pipeline execution. In two collaborations, with the US Department of Veterans Affairs and the US Food and Drug Administration, and on four open-source text and image data sets representative of other deployments, Snorkel provides 132% average improvements to predictive performance over prior heuristic approaches and comes within an average 3.60% of the predictive performance of large hand-curated training sets.