How machine learning models learn from training data

In machine learning, a model learns patterns from examples instead of following rules a programmer wrote by hand. In supervised learning, the model trains on labeled examples, such as thousands of emails marked "spam" or "not spam," and learns which features, like certain words or senders, tend to go with each label. A classification model sorts inputs into categories, like spam or not spam, while other models predict numbers, like tomorrow's temperature. To check whether a model really learned, developers test it on data it did not see during training; a model that memorizes its training data but performs poorly on new data is overfitting. A model can only be as good as its training data: if the data is too small, unrepresentative, or biased, for example a face-recognition system trained mostly on some groups of people, the model can be much less accurate for others. Improving the examples, adding more representative data, and choosing better features can make a model more accurate and fairer.

This explanation is AI-generated and can occasionally be wrong. 🚩 Report