What is data labeling?

Data labeling is essential in the context of supervised learning, where algorithms require labeled data to learn from. The process involves assigning tags or labels to data points, such as images, text, or audio, which helps the model understand what each piece of data represents. This allows the model to learn patterns and make predictions based on new, unseen data.

For instance, in a machine learning task involving image classification, each image might be labeled as "cat," "dog," or "bird." By providing these labels during the training phase, the algorithm can learn to differentiate between the categories and make accurate predictions when presented with new images.

The other options describe different processes related to data handling but do not specifically address the concept of assigning labels for the purpose of supervised model training. Annotation for unsupervised learning does not utilize labeled data, compression techniques focus on storage efficiency rather than labeling for learning purposes, and data cleaning or formatting pertains to preparing data for analysis without necessarily associating labels to it.