Weak Supervision

Full definition

The Snorkel framework (Ratner et al., Stanford, 2017) formalized the idea: domain experts write labeling functions (LFs) that vote noisily on labels, a generative model learns LF accuracies and correlations from agreement patterns, and a discriminative model trains on the resulting probabilistic labels. Weak supervision is now standard at Google (Snorkel Drybell), Apple, and most enterprise NLP teams. It scales labeling from human-bottlenecked to engineering-bottlenecked.

Why it matters

Weak supervision is how teams produce labeled datasets at the scale modern fine-tuning requires (>100k examples) without commensurate labeling spend. It is also the cleanest way to encode subject-matter expertise into training data.

Example

A telco classifies support tickets across 47 categories; 35 labeling functions written by subject-matter experts produce probabilistic labels for 4M tickets, training a model that outperforms one trained on 50k hand-labeled tickets.

Related terms

• Data Labeling
  Data labeling is the process of attaching ground-truth annotations to raw data — text, images, audio — so a supervised model can learn from it, ranging from yes/no classification to structured extraction to multi-turn preference comparisons.
• Active Learning
  Active learning is a labeling strategy in which the model selects which unlabeled examples a human should label next — typically the cases the current model is least certain about — focusing labeling effort on the most informative examples.
• Training Data
  Training data is the curated dataset used to teach a machine-learning model — every pattern the model can recognize, every bias it inherits, and every limit to its accuracy ultimately traces back to this data.
• Synthetic Data
  Synthetic data is artificially generated training data — produced by AI models or simulators rather than collected from the real world — used to augment, replace, or supplement real datasets while preserving privacy or covering rare cases.