What technique enhances training data by transforming or augmenting existing data?

The technique that enhances training data by transforming or augmenting existing data is data multiplication. This concept involves taking existing datasets and applying various transformations or augmentations to create new, synthetic data that helps improve model performance. Transformations can include geometric alterations, color modifications, noise addition, or other techniques that mimic real-world variations. This practice is especially important in fields like computer vision, where models benefit from exposure to a wide array of image conditions and styles. By multiplying the available data in this way, models can learn more effectively and generalize better to unseen data.

Other options mentioned, such as data privacy, focus on protecting sensitive information and ensuring compliance with regulations. Data preparation involves cleaning and organizing data for analysis but does not specifically refer to augmenting the dataset. Data quality management ensures that data is accurate and reliable but does not involve transformation or augmentation techniques.