Concept drift refers to a phenomenon in the practical application of AI in which some underlying statistics or characteristics of one or more variables changes after the deployment of a model such that a specific AI model's predictive accuracy changes. Concept drift is a problem that can be managed by periodically testing and updating the models or in some cases deploying models that take the possibility of concept drift into account.
- 1. Claude Sammut, Michael Harries. Concept Drift. (2020) doi:10.1007/978-0-387-30164-8_153
- 2. Big Data and Biomedical Informatics: A Challenging Opportunity. (2020) Yearbook of Medical Informatics. 23 (01): 08. doi:10.15265/IY-2014-0024 - Pubmed
- 3. Mirza B, Lin Z. Meta-cognitive online sequential extreme learning machine for imbalanced and concept-drifting data classification. (2016) Neural networks : the official journal of the International Neural Network Society. 80: 79-94. doi:10.1016/j.neunet.2016.04.008 - Pubmed