Publications

@inproceedings{9364223,

title = {A New Self Adaptive Fuzzy Unsupervised Clustering Ensemble Based On Spectral Clustering},

author = {Ines Lahmar and Aida Zaier and Mohamed Yahia and Ridha Bouallegue},

doi = {10.1109/SSD49366.2020.9364223},

issn = {2474-0446},

year  = {2020},

date = {2020-07-01},

booktitle = {2020 17th International Multi-Conference on Systems, Signals \& Devices (SSD)},

pages = {1-5},

abstract = {Nowadays, the goal of ensemble clustering approach, including the heterogeneity and huge number of variables, is to get the underlying structure based on the clustering quality results. In this context, we improve cluster ensemble approach based on cluster forest using Self-Adaptive Fuzzy C-Means (SAFCM) method to determine a good local base clustering. This proposed approach consists of two stages. To begin, unsupervised feature selection method is designed on the constructed of better variables on simulated data under the kappa metric. Next, we ameliorate the spectral clustering algorithm applied to find final grouping. Then, the SAFCM is used to obtain also the best number of K clusters and the optimal fuzzy exponent. This algorithm consists of three key steps: initially, a new algorithm is run which aims to get the best initial centroids values based on fuzzy density factor and distance. Next, new fuzzy cluster validity index (VI) is applied to evaluate the accurate number of the selected clusters. Finally, self adaptive FCM for the optimal value of Fuzzy weighting exponent m was proposed. To validate the effectiveness of the proposed clustering approach, we use eight datasets. Fur then, the performance of our system will be compared with several methods.},

keywords = {Performance evaluation;Adaptive systems;Clustering algorithms;Forestry;Feature extraction;Indexes;Feature Selection;Ensemble Clustering;Fuzzy system;Cluster validity;density;Spectral Clustering},

pubstate = {published},

tppubtype = {inproceedings}

}