Learning Algorithms

Learning algorithms are methods by which data is processed to extract patterns that can later be applied to novel situations. Generally, a system is said to learn if the performance of some task improves (with respect to a particular metric) after the analysis of data (experience). Familiar examples of learning systems are speech recognizers that adapt to individual users, automatic text translation services, and product recommenders. Underlying all of these techniques is a model, which defines what assumptions are made about the data and patterns that can be discovered in it. A learning algorithm is what is responsible for generating parameters for the model by processing collected data.