Supervised Enhanced Soft Subspace Clustering (SESSC) for TSK Fuzzy Classifiers
This work addresses the inefficiency of unsupervised clustering for TSK fuzzy classifiers in high-dimensional data, offering a domain-specific incremental improvement.
The paper tackled the problem of unsupervised clustering wasting label information in TSK fuzzy classifiers by proposing a supervised enhanced soft subspace clustering (SESSC) algorithm that integrates label information, within-cluster compactness, and between-cluster separation, resulting in improved performance on nine UCI datasets, especially with a small number of rules.
Fuzzy c-means based clustering algorithms are frequently used for Takagi-Sugeno-Kang (TSK) fuzzy classifier antecedent parameter estimation. One rule is initialized from each cluster. However, most of these clustering algorithms are unsupervised, which waste valuable label information in the training data. This paper proposes a supervised enhanced soft subspace clustering (SESSC) algorithm, which considers simultaneously the within-cluster compactness, between-cluster separation, and label information in clustering. It can effectively deal with high-dimensional data, be used as a classifier alone, or be integrated into a TSK fuzzy classifier to further improve its performance. Experiments on nine UCI datasets from various application domains demonstrated that SESSC based initialization outperformed other clustering approaches, especially when the number of rules is small.