Fuzzy Optimized Weighted based Locally Adaptive Clustering (FOWLAC) Algorithm in Benchmark Datasets

  • Vishal Goyal et al.

Abstract

In supervised classification, ensemble techniques are applied successfully. In data mining and recognition of pattern, the use of ensemble techniques is mostly attracted. It is an unsupervised clustering application. For specific and general tasks, various methods of clustering are proposed. Recent work introduced as Fuzzy Weighted Locally Adaptive Clustering (FWLAC) algorithm. Imbalanced data can be handled effectively by this proposed method.

There are two parameters that are manually tuned in this algorithm. These parameters effect the performance of FWLAC algorithm. Two solutions are proposed by this paper for well-tuning these parameters in Fuzzy Optimized Weighted based Locally Adaptive Clustering (FOWLAC) algorithm. Artificial Bee Colony (ABC) Algorithm is used to perform optimization of parameters. The pool of strategy is constructed using a Gbest-guided Artificial Bee Colony (GABC) and original ABC algorithms.

Best exploration results are shown by original ABC but it requires high time to converge. Multi-strategy Ensemble ABC (MEABC) algorithm is a stochastic algorithm which is based on weight. Bees that are generated randomly are used as an initial weights of the algorithm. Results of the FOWLAC and their existing methods are measured with respect to Accuracy, F-measure and Normalized Mutual Information (NMI). These metrics are used to validate a cluster results efficiently. The results of the clustering methods are experimented with breast cancer and bupa dataset.

Published
2019-12-12
How to Cite
et al., V. G. (2019). Fuzzy Optimized Weighted based Locally Adaptive Clustering (FOWLAC) Algorithm in Benchmark Datasets. International Journal of Control and Automation, 12(6), 99 - 107. Retrieved from http://sersc.org/journals/index.php/IJCA/article/view/2024