Instance-Based Learning: A Survey

Most classification methods are based on building a model in the training phase, and then using this model for specific test instances, during the actual classification phase. Thus, the classification process is usually a two-phase approach that is cleanly separated between processing training and test instances. As discussed in the introduction chapter of this book, these two phases are as follows: • Training Phase: In this phase, a model is constructed from the training instances. • Testing Phase: In this phase, the model is used to assign a label to an unlabeled test instance. Examples of models that are created during the first phase of training are decision trees, rule-based methods, neural networks, and support vector machines. Thus, the first phase creates pre-compiled abstractions or models for learning tasks. This is also referred to as eager learning, because the models are constructed in an eager way, without waiting for the test instance. In instance-based 157