Which term describes a model that fits training data very closely but generalizes poorly to new data?

Which term describes a model that fits training data very closely but generalizes poorly to new data?

• A. Underfitted Models
• B. Overfitted models
• C. Plateaus
• D. Vanishing Gradients

Answer: (B)

Overfitting is when a model learns the training data too well, capturing noise and peculiarities that don’t repeat in new data. This leads to very low training error but poor performance on unseen data because the model has effectively memorized the training set rather than learned general patterns. It often happens when the model has high capacity relative to the amount of data, allowing it to fit idiosyncrasies instead of underlying trends. In contrast, underfitting means the model is too simple to capture the patterns, resulting in error on both training and test data. Plateaus relate to slow learning in certain regions of the optimization landscape, and vanishing gradients describe diminishing gradient signals in deep networks—neither specifically describe poor generalization. So the term for a model that fits training data very closely but generalizes poorly is overfitted models.

Overfitting is when a model learns the training data too well, capturing noise and peculiarities that don’t repeat in new data. This leads to very low training error but poor performance on unseen data because the model has effectively memorized the training set rather than learned general patterns. It often happens when the model has high capacity relative to the amount of data, allowing it to fit idiosyncrasies instead of underlying trends. In contrast, underfitting means the model is too simple to capture the patterns, resulting in error on both training and test data. Plateaus relate to slow learning in certain regions of the optimization landscape, and vanishing gradients describe diminishing gradient signals in deep networks—neither specifically describe poor generalization. So the term for a model that fits training data very closely but generalizes poorly is overfitted models.