Concept drift

How it is used

Used in model monitoring. It differs from an input distribution shift that leaves the input-outcome relationship unchanged.

Example sentences

1. We observed concept drift when the same shopping behavior began predicting a different outcome.

   The meaning of a behavioral signal changed, not merely how often that signal appeared.

2. The fraud team investigated concept drift after attackers changed their transaction patterns.

   Previously useful input patterns may no longer distinguish fraud in the same way.

Common collocations

• detect concept drift
• concept drift monitoring

Related terms

• data driftA change over time in the data a deployed model receives, compared with the data it was trained on, which can slowly make its predictions worse.
• model evaluationThe process of measuring how well a machine learning model performs on chosen data and metrics after training.

More terms: Machine Learning and Data Processing

• artificial intelligence
• machine learning
• model
• dataset
• training data
• label
• train
• supervised learning
• neural network
• deep learning
• loss function
• overfitting