End-to-End Weak Supervision

Abstract

Aggregating multiple sources of weak supervision (WS) can ease the data-labeling bottleneck prevalent in many machine learning applications, by replacing the tedious manual collection of ground truth labels. Current state of the art approaches that do not use any labeled training data, however, require two separate modeling steps: Learning a probabilistic latent variable model based on the WS sources -- making assumptions that rarely hold in practice -- followed by downstream model training. Importantly, the first step of modeling does not consider the performance of the downstream model. To address these caveats we propose an end-to-end approach for directly learning the downstream model by maximizing its agreement with probabilistic labels generated by reparameterizing previous probabilistic posteriors with a neural network. Our results show improved performance over prior work in terms of end model performance on downstream test sets, as well as in terms of improved robustness to dependencies among weak supervision sources.

arXiv commentCode URL: https://github.com/autonlab/weasel

References in corpus (9)

• Snorkel: Rapid Training Data Creation with Weak Supervision
• Learning the Structure of Generative Models without Labeled Data
• Learning From Noisy Singly-labeled Data
• Inferring Generative Model Structure with Static Analysis
• Learning from Rules Generalizing Labeled Exemplars
• Learning Dependency Structures for Weak Supervision Models
• Data Programming using Continuous and Quality-Guided Labeling Functions
• Dependency Structure Misspecification in Multi-Source Weak Supervision Models
• Comparing the Value of Labeled and Unlabeled Data in Method-of-Moments Latent Variable Estimation