Glossary

Overfitting is a problem that arises in statistical modelling, machine learning, and artificial intelligence when a given model is overly adjusted to specific examples in the training data and is consequently no longer capable of making general predictions regarding new, unfamiliar data. Overfitting occurs when a model’s complexity is too great in relation to the available training data, or when the model – due to repetitions of the training that are too long or too numerous – inadvertently learns the training data “by heart.”