Definition
The process of using a fully trained machine learning model to make a prediction or generate an output when presented with new, unseen input data.
In short: The process of using a fully trained machine learning model to make a prediction or generate an output when presented with new, unseen input data.