Training & Learning
Overfitting
When a model memorizes its training data too closely and then performs badly on anything new.
Overfitting happens when a model learns the quirks and noise of its training set instead of the general pattern, so it looks excellent on familiar data and fails on fresh data. It is one of the most common ways machine learning projects go wrong. The usual defenses are more and more varied data, simpler models, and techniques that discourage the model from leaning too hard on any single detail.
Reference: Wikipedia: Overfitting
Related terms
Underfitting
When a model is too simple or undertrained to capture the real pattern, so it performs poorly even on its training data.
Training Data
The examples a model learns from, whose size, quality, and bias largely determine how the model turns out.
Loss Function
The formula that scores how wrong a model is, giving training a single number to push downward.
Epoch
One complete pass of the training process through the entire training dataset.