Noise Filtering

This paper continues to explore noise filtering using learners. Four classification scenarios were investigated. The first scenario applies the more classical approach: training one classifier with a single fit dataset and predicting the test dataset. The second approach is a popular method in data mining: a classifier is built based on the prediction of multiple learners induced on the same dataset. The third approach consists of using the prediction of the same learner induced on multiple fit datasets (multi-dataset classifier). Finally, the most generic approach combines the predictions of multiple learners built on multiple fit datasets and applied on the dataset we want to predict. Such a technique is referred to as multi-learner multi-dataset classifier.

To our knowledge, this empirical work is one of the largest in terms of both scale and scope: 119 (17 9 7) base classification models were built, and more than 700 vectors of base estimates were generated. This paper was published in the Software Quality Journal. You can find more information on Springer and on the ACM portal.

Dr. Khoshgoftaar and I outline our two filtering techniques, the multiple-partitioning filter and the iterative-partitioning filter respectively. The primary aim of this study is to compare the predictive performances of the final models built on the filtered and the un-filtered training datasets. A case study of software measurement data provided by the NASA is performed. The data is available through the Metrics Data Program (MDP) and includes software measurement data and associated error data collected at the subroutine level.

This article was published in the Journal of Computer Science and Technology (JCST) in Volume 22, Issue 3 in 2006. Details of the paper can be found on the editor web site.

This is the first paper where Dr. Khoshgoftaar and I present the Ensemble Filter to filter noise using machine learners. We show that this filter can be specialized into the classification, ensemble, multiple-partitioning, or iterative-partitioning filter. A case study of software metrics data from a high assurance software project analyzes the similarities between the filters obtained from the specialization of the Ensemble Filter.

This paper was published part of the Proceedings of the 2004 IEEE International Conference on Information Reuse and Integration (IRI04). You can find more information on IEEE.