Improving Automated Land Cover Mapping by Identifying and Eliminating Mislabeled Observations from Training Data

This paper presents a new approach to identifying and eliminating mislabeled training samples. The goal of this technique is to decrease the error of classification algorithms by improving the quality of the training data. The approach employs an ensemble of classifiers that serve as a filter for the training data. Using an n-fold cross validation, the training data is passed through the filter. Only samples that the filter classifies correctly are passed to the final classification algorithm. An empirical evaluation of the approach on the task of automated land cover mapping illustrates that the ensemble filter approach is an effective method for identifying labeling errors.