| Multi-objective Ensemble Construction, Learning and Evolution (2008) | |||||||||||||||
Abstract | |||||||||||||||
| Abstract. An ensemble of learning machines has been theoretically and empirically shown to generalise better than single learners. Diversity and accuracy are two key properties that ensemble members should possess in order for this generalisation principle to hold. Viewing these properties as objectives, we take the position of rendering multi-objective evolutionary algorithms as effective solution concepts to the problem of ensemble construction and learning. 1 | |||||||||||||||
Details der Publikation | |||||||||||||||
| |||||||||||||||