Baolei Li

Data clustering using multivariant optimization algorithm

Data clustering is one of the most popular techniques in data mining to group data with great similarity and high dissimilarity into each cluster. This paper presents a new clustering method based on a novel heuristic optimization algorithm proposed recently and named as multivariant optimization algorithm (MOA) to locate the optimal solution automatically through global and local alternating search implemented by a global exploration group and several local exploitation groups. In order to demonstrate the performance of MOA-clustering method, it is applied to group six real-life datasets to obtain their clustering results, which may be compared with those received by employing K-means algorithm, genetic algorithm and particle swarm optimization. The results show that the proposed clustering algorithm is an effective and feasible method to reach a high accurate rate and stability in clustering problems.

On the convergence of multivariant optimization algorithm

Graphical abstractDisplay Omitted HighlightsMultivariant optimization algorithm is proposed to solve complex problems efficiently.The algorithm is proved to be globally convergent by theoretical analysis.Experimental results show MOAs strengths in convergence accuracy and probability. In this paper, we introduce a new global optimization method and study its global convergence property through theoretical and experimental approaches. The proposed method is named as multivariant optimization algorithm (MOA) because the intelligent searchers, which are called as atoms, not only are divided into multiple subgroups but also are variant in responsibility. That is, global atoms explore the whole solution space in the hope of finding potential areas where local atoms start the local exploitation. The proposed method is characterized by two important features. On one hand, global atoms do the global exploration in each loop to jump out from local traps. On the other hand, global and local atoms conduct the global exploration and the local exploitation according to their own responsibility, respectively. These features contribute to increasing the chance of converging to the global best. To study the convergence property of MOA, we carried out the convergence analysis, numerical optimization experiments and the shortest path planning experiments. And the results demonstrate that MOA is globally convergent and superior to the compared methods in the global convergence accuracy and probability in solving complex challenging problems which have one or more features such as deceptiveness, randomly located optimum, asymmetry or multiple traps.

Classification of diffuse large B cell lymphoma gene expression data based on two-layer particle swarm optimization

Classification of gene expression data to determine subtype of samples is meaningful to research tumors in molecular biology level. It is also an important way to make further treatment plan for the patient. Particle swarm optimization (PSO) is proven to be an ineffective solution for classification and clustering in bioinformatics as it could not give a stable prediction result. In this study, a classifier based on the two layer particle swarm optimization (TLPSO) algorithm and uncertain training sample sets is established. Samples of diffuse large B cell lymphoma (DLBLC) gene expression data are used for training and validating. The classification stability and accuracy by the proposed TLPSO algorithm increase significantly compared with the results obtained by using algorithms known as PSO and K-means.

A family Particle Swarm Optimization based on the family tree

The concept of the family was previously introduced into Particle Swarm Optimization (PSO). To further study the multi-group structure of the Family PSO (FPSO), this paper introduces the family tree into the FPSO. It made different families form a family tree and a swarm consisted of some family trees. In the experiment, topological distance was used to form a family and three spatial structures of the family trees were simply defined and demonstrated in two-dimensional space. Simulations for seven benchmark functions demonstrated that two family trees in a swarm and each family had 2 particles that performed better than other combinations. Results also showed the multi-group structure of FPSO was a problem deserving of study at high-dimensional space.

On the unbiasedness of Multivariant Optimization Algorithm

Multivariant Optimization Algorithm (MOA) is proposed to effectively solve complex multimodal optimization problems through tracking the history information by multiple variant search groups based on a structure. The proposed method has the ability to locate optimum through global-local search iterations which are carried out by a global exploration group and local exploitation groups which are not only multiple but also variant. In this paper, we study the unbiasedness property of MOA and prove that MOA provides an unbiased estimate of the optimal solution for identification problem on an AR model where the outputs are corrupted by noises. The comparison experiments on the identifications of AR model by (Finite Impulse Response) FIR filter shows that MOA is superior to recursive least squares (RLS) and the particle swarm optimization (PSO) in unbiasedness property.

The imperfect sample points of interpolation based on B-spline suface reconstruction

A novel B-spline surface interpolation algorithm based on particle swarm optimization is proposed to solve surface optimization problem. The core of this paper is that studying the effect of missing sample point data about B-spline surfaces interpolation. First, given sample points are used to interpolate B-spline surfaces, and then after removing some sample points, the remaining sample points are applied to generate a new B-spline surface. Finally, the image and error in previous process are analyzed.

Recursive particle swarm optimization applications in radial basis function networks modeling system

A novel strategy on particle swarm optimization is proposed to solve dynamic optimization problems, in which the data are obtained not once for all but one by one. The evolutionary states of the particle swarm are guided recursively by the proposed algorithm, according to the information achieved by the continuous data and the prior estimated knowledge on the solution space. The experimental results for three test functions show that radial basis function networks modeling system based on the proposed recursive algorithm requires fewer radial basis functions and gives more accurate results than other traditional improved PSO in solving dynamic problems.

Classification of leukemia gene expression profiles based on multivariant optimization algorithm

Classification of leukemia samples based on gene expression profiles has been proved an efficient way. Large numbers of intelligence algorithms have been exploited based on this purpose. However, few of them display stable and accurate performance for both low and high gene dimensionalities. Still none of them could keep the history information of optimization. Here, a classification algorithm based on the novel multivariant optimization algorithm (MOA) is proposed. Leukemia gene expression profiles with different dimensionalities are used for validation. The particle swarm optimization (PSO) and the two-layer particle swarm optimization (TLPSO) algorithm are used for comparison. The MOA shows stable and relatively accurate classification performance and could be used as an effective classification algorithm for gene expression profiles.

A novel multivariant optimization algorithm for multimodal optimization

This paper provides a detailed description of a novel multivariant optimization algorithm (MOA) for multi-modal optimization with the main idea to share search information by organizing all search atoms into a special designed structure. Its multiple and variant group property make MOA capable on multi-modal optimization problems. The capability of the MOA method in locating and maintaining multi optima in one execution is discussed in details in this paper and two experiments are carried out to validate its feasibility in multi-modal optimization problems. The experimental results are also compared with those obtained by the species-based PSO, the adaptive sequential niche PSO and the memetic PSO. The experiment results show that MOA has high success rate and convergence speed in multi-modal optimization problems.

The global double cubic B-spline surface interpolation based on particle swarm optimization

a novel B-spline surface interpolation algorithm based on particle swarm optimization is proposed to solve surface optimization problem. Two steps of the algorithm are established in present paper: first, control points are calculated by a given cloud of 3D data points. Second, a set of optimal parameters of the data points is obtained by using particle swarm optimization. The results are compared with the sample points to minimize the root square error. Compared to the traditional method with the smallest root square error, the experimental results have shown that particle swarm optimization yields better solution.