Meta-Learning of Instance Selection for Data Summarization

The goal of instance selection is to identify which instances (examples, patterns) in a large dataset should be selected as representatives of the entire dataset, without significant loss of information. When a machine learning method is applied to the reduced dataset, the accuracy of the model should not be significantly worse than if the same method were applied to the entire dataset. The reducibility of any dataset, and hence the success of instance selection methods, surely depends on the characteristics of the dataset. However the relationship between data characteristics and the reducibility achieved by instance selection methods has not been extensively tested. This chapter adopts a meta-learning approach, via an empirical study of 112 classification datasets, to explore the relationship between data characteristics and the success of a naive instance selection method. The approach can be readily extended to explore how the data characteristics influence the performance of many more sophisticated instance selection methods.