overfit

/ˌoʊvərˈfɪt/
verb
  1. To design a model (especially in statistics or machine learning) that matches the training data too closely, so it performs poorly on new data.
    • If you train a neural network for too long, you may overfit the training data.
    • To avoid overfitting, data scientists use techniques like cross-validation.
    • The algorithm overfit the noise in the dataset instead of learning the real pattern.
Antonyms