During training, a model produces outputs that are compared with target values or an objective function. An optimization algorithm adjusts parameters to reduce the measured error or improve a defined reward. Evaluation on data not used for training helps estimate generalization and identify overfitting.
How Machine Learning Models Learn from Data
Machine-learning models adjust internal parameters to capture patterns in training data.
Explanation
Topics
Sources and further reading
Sources are listed as supplied in the Scienexy dataset. A listed source is not, by itself, a guarantee that every sentence has been independently fact-checked. Estimates and scientific interpretations may depend on methods, context and newer evidence.