A new approach for semi-supervised clustering based on Fuzzy C-Means

Abstract

In traditional machine learning applications, only labeled data is used to train the classifier. Labeled data are difficult, expensive, time-consuming and require human experts to be obtained in several real applications. Semi-supervised learning address this issue. Semi-supervised learning uses large amount of unlabeled data, combined with the labeled data, to build better classifiers. The semi-supervised algorithm could be an extension of an unsupervised algorithm. Such algorithm would be based on unsupervised clustering algorithms, adding a term in its objective function that makes use of labeled information to guide the learning process. This study presents a new algorithm for semi-supervised clustering based on Fuzzy C-Means algorithm. The classifier was evaluated and compared against two semi-supervised clustering algorithms in the context of learning from partially labeled data. The behavior of the proposed algorithm is discussed and the results are validated using cross-validation and the confidence interval. Thus, it was possible to certify the better accuracy performance of the new algorithm when a few labeled data are available.978-1-4244-8126-2/10/$26.00 ©2010 IEEE

An adaptive semi-supervised fuzzy clustering algorithm based on objective function optimization

“…More relevant to our work are the approaches investigated by Bouchachia and Pedrycz [7] and Macario and De Carvalho [8] which are typical examples of objective function optimization models for clustering with partial supervision. The algorithms basically extends the objective function of the Fuzzy C-Means (FCM) [13] algorithm.…”

“…The original algorithm proposed in Macario and De Carvalho [8] work can be formulated setting λ k = (λ k1 , . .…”

“…Enjoying this information, the algorithm optimizes the objective function to label examples in a precise way. The developed algorithm proposed in the Macario and De Carvalho works [8] adds a second term supervised to the original objective function of Fuzzy CMeans algorithm [13]. We developed an optimization function where the data points are compared, and if they belong to the same class, the difference between membership degrees to the cluster that they belong are attenuated.…”

“…In especial, the interest of this work is in semi supervised clustering with function optimization techniques, due the possibility of incorporating labeled data in classical unsupervised clustering algorithms [6], [7], [8]. This paper extends the clustering semi-supervised algorithm presented by Macario and De Carvalho [8] incorporating an adaptive distance approach allowing the construction of fuzzy partitions in ellipsoids format.…”

“…This paper extends the clustering semi-supervised algorithm presented by Macario and De Carvalho [8] incorporating an adaptive distance approach allowing the construction of fuzzy partitions in ellipsoids format.…”