ABSTRACT
Since the late 1950s, computer scientists (particularly those working in the area of artificial intelligence) have been trying to understand how to construct computer programs that perform tasks we normally think of as requiring human intelligence, and which can improve their performance over time by modifying their behavior in response to experience. In other words, one objective has been to design computer programs that can learn. For example, Samuels designed a program to play checkers in the early 1960s that could improve its performance as it gained experience playing against human opponents. More recently, research on artificial neural networks has stimulated interest in the design of systems capable of performing tasks that are difficult to describe algorithmically (such as recognizing a spoken word or identifying an object in a complex scene), by exposure to many examples. As a concrete example consider the task of handwritten character recognition. A learning
algorithm is given a set of examples where each contains a handwritten character as specified by a set of attributes (e.g., the height of the letter) along with the name (label) for the intended character. This set of examples is often called the training data. The goal of the learner is to efficiently construct a rule (often referred to as a hypothesis or classifier) that can be used to take some previously unseen character and with high accuracy determine the proper label.
