Meta-Learning in Computational Intelligence

This book defines and reveals new theoretical and practical trends in meta-learning, inspiring the readers to further research in this exciting field.

Abstract

In this chapter, we provide a survey of the various architectures that have been developed, or simply proposed, to build extended meta-learning systems that cover entire data mining workflows. They all consist of integrated repositories of meta-knowledge on the knowledge discovery process and leverage that information to propose useful workflows. Our main observation is that most of these systems are very different, and were seemingly developed independently from each other, without really capitalizing on the benefits of prior systems. By bringing these different architectures together and highlighting their strengths and weaknesses, we aim to reuse what we have learned, and we draw a roadmap towards a new generation of knowledge discovery support systems.