Selective Sampling
Selective Sampling is a data collection approach in which specific samples are intentionally chosen based on predefined criteria, rather than selecting observations entirely at random.
What is Selective Sampling?
Selective sampling allows a machine learning system or data collection process to prioritize particular examples based on factors such as uncertainty, rarity, relevance, or expected information value. In active learning, for example, a model may select uncertain samples for labeling instead of labeling the entire dataset.
Why is Selective Sampling Important?
Selective sampling can reduce the cost and effort required to collect or label data while focusing attention on the most informative examples. However, poorly designed selection criteria can introduce sampling bias and make the resulting dataset less representative.
Common use cases
Selective Sampling is commonly used in active learning, dataset creation, human labeling workflows, rare-event detection, model evaluation, and machine learning research.