Identifying Mislabeled Training Data

C. E. Brodley, M. A. Friedl

This paper presents a new approach to identifying and eliminating mislabeled training instances for supervised learning. The goal of this approach is to improve classification accu…