Whenever a learner finds a pattern in the data that is not actually true in the real world, we say that it has overfit the data. Overfitting is the central problem in machine learning. More papers have been written about it than about any other topic.
