Ferani E. Zulvia

Automatic kernel clustering with bee colony optimization algorithm

Cluster analysis is important in data mining, especially if there is unsupervised data. Recently, many clustering methods have been proposed. Unfortunately, most of these require the definition of the number of clusters, in advance. This study addresses this weakness by proposing a new automatic clustering algorithm: automatic kernel clustering with bee colony optimization (AKC-BCO). AKC-BCO determines the appropriate number of clusters and assigns data points to correct clusters. This is accomplished by the kernel function, which increases clustering capability. This method is validated using several benchmark data sets. The result is compared with several existing automatic clustering methods. The experiment results demonstrate that the proposed AKC-BCO is more stable and accurate than others. Furthermore, the proposed method is also applied to a real-world medical problem.

The gradient evolution algorithm

This study presents a new metaheuristic method that is derived from the gradient-based search method. In an exact optimization method, the gradient is used to find extreme points, as well as the optimal point. This study modifies a gradient method, and creates a metaheuristic method that uses a gradient theorem as its basic updating rule. This new method, named gradient evolution, explores the search space using a set of vectors and includes three major operators: vector updating, jumping and refreshing. Vector updating is the main updating rule in gradient evolution. The search direction is determined using the Newton-Raphson method. Vector jumping and refreshing enable this method to avoid local optima. In order to evaluate the performance of the gradient evolution method, three different experiments are conducted, using fifteen test functions. The first experiment determines the influence of parameter settings on the result. It also determines the best parameter setting. There follows a comparison between the basic and improved metaheuristic methods. The experimental results show that gradient evolution performs better than, or as well as, other methods, such as particle swarm optimization, differential evolution, an artificial bee colony and continuous genetic algorithm, for most of the benchmark problems tested.

An application of a metaheuristic algorithm-based clustering ensemble method to APP customer segmentation

This study proposes a metaheuristic-based clustering ensemble method. It integrates the clustering ensembles algorithm with the metaheuristic-based clustering algorithm. In the clustering ensembles, this study performs an improved generation mechanism and a co-association matrix in the co-occurrence approach. In order to improve the efficiency, a principle component analysis is employed. Furthermore, three metaheuristic-based clustering algorithms are proposed. This paper uses a real-coded genetic algorithm, a particle swarm optimization and an artificial bee colony optimization to combine with clustering ensembles algorithm. The experimental results indicate that the proposed metaheuristic-based clustering ensembles algorithms have better performance than metaheuristic-based clustering without clustering ensembles method. Furthermore, the proposed algorithms are applied to solve a customer segmentation problem. The real problem is come from a mobile application. Among all of the proposed algorithms, the artificial bee colony optimization-based clustering ensembles algorithm outperforms other algorithms. Therefore, the marketing strategy for the real application is made based on the best result.

Solving bi-level linear programming problem through hybrid of immune genetic algorithm and particle swarm optimization algorithm

Bi-level linear programming, consisting of upper level and lower level objectives, is a technique for modeling decentralized decision. This study presents a hybrid of immune genetic algorithm and vector-controlled particle swarm optimization (IGVPSO) to solve the bi-level linear programming problem (BLPP). It is applied to a supply chain model that is a BLPP. Using four problems from the literature and the supply chain distribution models, the computational results indicate that the proposed method is superior to some algorithms.

Solving CVRP with time window, fuzzy travel time and demand via a hybrid ant colony optimization and genetic algortihm

This study intends to propose a hybrid ant colony optimization (ACO) and genetic algorithm (GA) (HACOGA) for solving the capacitated vehicle routing problem (CVRP) with time window, fuzzy travel time and demand. A mathematical model for CVRP with time window, fuzzy travel time and demand is first constructed. It applies fuzzy credibility and ranking approaches. Then, the proposed HACOGA which combines ACO with GA to accelerate its exploration is employed. It also embeds local search algorithms to generate a better initial solution and improve its performance at the end of evolution. The proposed algorithm is verified using an instance of CVRP with time window and fuzzy travel time first. The simulation result indicates that the proposed HACOGA outperforms previous methods. Furthermore, a simulation example is employed to show the effectiveness of the proposed algorithm for solving CVRP with time window, fuzzy travel time and fuzzy demand. The computational results reveal that HACOGA still has the best performance.

Hybrid genetic ant colony optimization algorithm for capacitated vehicle routing problem with fuzzy demand — A case study on garbage collection system

The logistics distribution problem, which is typically represented as the capacitated vehicle routing problem (CVRP), is a very important issue in logistics. Thus, this paper considers the CVRP with fuzzy demand (CVRPFD). As an NP-hard problem, many researches in CVRP apply meta-heuristic method rather than exact method. A hybrid genetic algorithm (GA) and ant colony optimization (ACO) is proposed in this study. It combines advantages of GA, ACO and two local search methods, namely Prims algorithm and 2-opt. Verification of the proposed methods performance is conducted on eight benchmark data sets in CVRP. The results show that the proposed GACO algorithm is competitive with other existing algorithms for solving CVRP. Furthermore, the proposed method is also applied to solve garbage collection system which is represented as CVRPFD. The results are also very promising.

Cluster analysis using a gradient evolution-based k-means algorithm

Cluster analysis is a very useful data analysis tool. It can reveal the hidden information stored inside a dataset. Therefore, many researches proposed different clustering algorithms. This paper intends to propose a gradient evolutionbased Ä-means algorithm. Ä-means algorithm is a well-known clustering algorithm. It offers a simple algorithm to divide the dataset into several clusters. Unfortunately, its results are highly influenced by the initial centroids. Unpromising initial centroid might lead the k-means to the bad clustering result. This paper aims to improve this drawback by adopting a new metaheuristic algorithm, named a gradient evolution (GE) algorithm. In this paper, we proposed a GE-based Ä-means algorithm for solving the clustering problems. The proposed algorithm is validated by using some benchmark datasets. The computation results showed that the proposed algorithm can obtain better results compared with some other metaheuristic-based k-means algorithms.