hmu.ai
Back to AI Dictionary
AI Dictionary

Active Learning

Definition

A training process where the model queries a user to label data points that are most informative.

Deep Dive

Active learning is a machine learning paradigm where the learning algorithm is able to interactively query a user or another information source to obtain the desired outputs (labels) for new data points. Unlike traditional supervised learning, where the model passively receives a fixed, labeled dataset, an active learner intelligently selects the most informative, unlabeled data points to be labeled. This strategy is particularly valuable in scenarios where unlabeled data is abundant but labeling data is expensive, time-consuming, or requires domain expertise.

Examples & Use Cases

  • 1An image classification system asking a human expert to label a small subset of images that it finds most ambiguous or difficult to classify
  • 2A medical diagnosis AI querying a doctor for a definitive diagnosis on specific patient cases where its confidence score is low
  • 3A sentiment analysis model asking a human to categorize review comments it finds hard to definitively label as positive, negative, or neutral

Related Terms

Semi-supervised LearningData LabelingHuman-in-the-Loop

Part of the hmu.ai extensive business and technology library.