A two-leveled symbiotic evolutionary algorithm for clustering problems

Abstract

Because of its unsupervised nature, clustering is one of the most challenging problems, considered as a NPhard grouping problem. Recently, several evolutionary algorithms (EAs) for clustering problems have been presented because of their efficiency for solving the NP-hard problems with high degree of complexity. Most previous EA-based algorithms, however, have dealt with the clustering problems given the number of clusters (K) in advance. Although some researchers have suggested the EA-based algorithms for unknown K clustering, they still have some drawbacks to search efficiently due to their huge search space. This paper proposes the two-leveled symbiotic evolutionary clustering algorithm (TSECA), which is a variant of coevolutionary algorithm for unknown K clustering problems. The clustering problems considered in this paper can be divided into two sub-problems: finding the number of clusters and grouping the data into these clusters. The two-leveled framework of TSECA and genetic elements suitable for each sub-problem are proposed. In addition, a neighborhood-based evolutionary strategy is employed to maintain the population diversity. The performance of the proposed algorithm is compared with some popular evolutionary algorithms using the real-life and simulated synthetic data sets. Experimental results show that TSECA produces more compact clusters as well as the accurate number of clusters.

A Novel Effective Distance Measure and a Relevant Algorithm for Optimizing the Initial Cluster Centroids of K-means

“…Literature [22], [23], [24], [25], [26], [27] propose that deep belief networks can be used to downscale and classify data, and that input data can be progressively feature extracted from low to high levels to improve the accuracy of classification. The literature [28], [29], [30] raised a greedy unsupervised training method [31], [32], [33], [34] which divides the learning process of a DBN into two steps, unsupervised learning by extracting input information layer by layer and supervised learning by fine-tuning the entire network with fixed labels, solving the problem of the difficulty of learning multiple implicit layer parameters. Some literature has conducted more in-depth research on both deep confidence network-based methods and deep learning methods based on active learning [35], [36] and proposed a series of deep confidence network-based classification methods to improve classification performance.…”

An evolutionary algorithm with the partial sequential forward floating search mutation for large-scale feature selection problems

“…Garcı´a et al (2006) presented a parallel scatter search meta-heuristic algorithm in which a set of solutions evolves based on the mechanisms of combination between solutions. JirapechUmpai and Aitke (2005) applied an EA to microarray classification in order to search for the optimal or nearoptimal set of predictive genes, while Shin et al (2012) proposed a two-level symbiotic EA for high-dimensional clustering problems. In addition, Oh et al (2004) proposed a hybrid GA for feature selection, which embeds an SFFS local search operation into the conventional GA, leading to significant improvement in the final performance.…”

“…EVALUATE new candidates; 5. SELECT individuals for the next generation; UNTIL (Termination Condition is satisfied) END The limitations of pure meta-heuristic algorithms such as EA and GA have been uncovered in many applications (Oh et al, 2004;Shin et al, 2012). An effective way to overcome a drawback of conventional GA, or fine-tuning near local optimum points with long running times, is to hybridize GA by embedding the SFFS local search operations (Oh et al, 2004).…”

Hierarchical Clustering of GaussianMixture Models in Applications forContinuous Speech Recognition

“…Користе се и у процесу кластеровања података, као и за естимацију густине вероватноћа. Процедура кластеровања представља једну од најзначајнијих компоненти било ког задатка препознавања [4]- [6]. Хијерархијско кластеровање модела Гаусових смеша [7] [8] (Hierarchical Clustering of Gaussian Mixture Models -HGMMC) кључна је компонента метода селекције гаусијана [9]- [12], који се примењује за повећање брзине система за препознавање говора (говорника или емоција) базираном на моделима Гаусових смеша, и то посебно у апликацијама за континуално препознавање говора (Continuous Speech Recognition -CSR), верификацију гласом [3] и адаптацију на говорника [13].…”