In machine learning, what is the distinction between training and inference?

• A. Training teaches a model using a dataset, while inference applies the trained model to make predictions on new data.
• B. Training and inference both apply a trained model to new data.
• C. Inference teaches a model using a dataset; training applies to new data.
• D. Training is about evaluating model performance on a test set only.

Answer: (A)

Training and inference refer to two distinct phases in machine learning. Training is the learning phase: the model uses labeled data to adjust its internal parameters so its predictions align with the true labels, aiming to minimize error. Inference is the deployment phase: the already trained model is used to make predictions on new, unseen data without changing its learned parameters.

For example, you train a spam classifier on a labeled set of emails, adjusting the model to distinguish spam from legitimate messages. Once trained, you deploy it to classify incoming emails in real time. The distinction is clear: training teaches the model from data, while inference applies the trained model to new data to generate predictions.

This helps separate the learning process from the act of making predictions.