Ranking of Classifiers Using Active Meta Learning

In Classification, Model Selection is one of the critical issues as different models from different categories are available. To select the best model for any given data set is a challenging task. Meta Learning automates this task by acquiring knowledge from the past experience and stores this knowledge into database called Meta Knowledge Base. When new data set comes, stored knowledge can be used for proving ranking of the candidate algorithms. But one of the problems with Meta Learning is generation of Meta Examples as large number of candidate algorithms and data sets are available. To reduce the generation of Meta Examples into Meta Knowledge Base, Active Meta Learning can be used that reduces generation of Meta Examples and at the same time maintaining the performance of candidate algorithms. In this book, Ranking is provided using Active Meta Learning approach by considering Data set Characteristics.