End-to-End Weak Supervision

TL;DR

This paper introduces an end-to-end method for weak supervision that directly optimizes the downstream model, leading to better performance and robustness compared to traditional two-step approaches.

Contribution

It proposes a novel end-to-end framework that integrates weak source modeling with downstream model training, bypassing the limitations of separate probabilistic modeling.

Findings

Improved downstream test set performance.

Enhanced robustness to source dependencies.

Outperforms prior weak supervision methods.

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.