Weighted Particle Swarm Clustering Algorithm for Self-Organizing Maps

The prediction of high quality soil moisture is of great significance to agricultural production and scientific research. In order to solve the problem that the prediction results of (ELM) regression model of limit learning machine are affected by input parameters, the stochastic weight particle swarm optimization algorithm (RandWPSO) is applied to ELM regression model. In this paper, a soil moisture prediction method of particle swarm optimization limit learning machine based on random inertia weight is proposed. In this method, the data of soil temperature and light intensity measured by sensor are used to preprocess the data, the training sample set is constructed, and the ELM regression model is established. The input weight and threshold in ELM are optimized by using random weight particle swarm optimization algorithm to avoid falling into local optimization, thus the prediction model of soil moisture based on RandWPSO-ELM is established. The soil moisture of sugar beet in Hulan area was studied. The experimental results show that the method has high accuracy and stability, and can provide an effective reference for the growth of sugar beet in greenhouse.

Extending the Kohonen self-organizing map networks for clustering analysis

A fuzzy self-organizing map (SOM) network is proposed in this paper for visual motor control of a 7 degrees of freedom (DOF) robot manipulator. The inverse kinematic map from the image plane to joint angle space of a redundant manipulator is highly nonlinear and ill-posed in the sense that a typical end-effector position is associated with several joint angle vectors. In the proposed approach, the robot workspace in image plane is discretized into a number of fuzzy regions whose center locations and fuzzy membership values are determined using a Fuzzy C-Mean (FCM) clustering algorithm. SOM network then learns the inverse kinematics by on-line by associating a local linear map for each cluster. A novel learning algorithm has been proposed to make the robot manipulator to reach a target position. Any arbitrary level of accuracy can be achieved with a number of fine movements of the manipulator tip. These fine movements depend on the error between the target position and the current manipulator position. In particular, the fuzzy model is found to be better as compared to Kohonen self-organizing map (KSOM) based learning scheme proposed for visual motor control. Like existing KSOM learning schemes, the proposed scheme leads to a unique inverse kinematic solution even for a redundant manipulator. The proposed algorithms have been successfully implemented in real-time on a 7 DOF PowerCube robot manipulator, and results are found to concur with the theoretical findings.

Cable-driven parallel robots (CDPRs) have been widely used in engineering fields because of their significant advantages including high load-bearing capacity, large workspace, and low inertia. However, the impact of convergence speed and solution accuracy of optimization approaches on optimal performances can become a key issue when it comes to the optimal design of CDPR applied to large storage space. An adaptive adjustment inertia weight particle swarm optimization (AAIWPSO) algorithm is proposed for the multi-objective optimal design of CDPR. The kinematic and static models of CDPR are established based on the principle of virtual work. Subsequently, two performance indices including workspace and dexterity are derived. A multi-objective optimization model is established based on performance indices. The AAIWPSO algorithm introduces an adaptive adjustment inertia weight to improve the convergence efficiency and accuracy of traditional particle swarm optimization (PSO) algorithm. Numerical examples demonstrate that final convergence values of the objective function by the AAIWPSO algorithm can almost be 14∼20% and 19∼40% higher than those by the PSO algorithm and genetic algorithm (GA) for the optimal design of CDPR with different configurations and masses of end-effectors, respectively.

This paper proposes a clustering ensemble method that introduces cascade structure into the self-organizing map (SOM) to solve the problem of the poor performance of a single clusterer. Cascaded SOM is an extension of classical SOM combined with the cascaded structure. The method combines the outputs of multiple SOM networks in a cascaded manner using them as an input to another SOM network. It also utilizes the characteristic of high-dimensional data insensitivity to changes in the values of a small number of dimensions to achieve the effect of ignoring part of the SOM network error output. Since the initial parameters of the SOM network and the sample training order are randomly generated, the model does not need to provide different training samples for each SOM network to generate a differentiated SOM clusterer. After testing on several classical datasets, the experimental results show that the model can effectively improve the accuracy of pattern recognition by 4%∼10%.

Density Peaks Clustering (DPC) algorithm identifies cluster centers based on local density and relative distance, disregarding the effect of the sample's environment on its density. This makes it challenging to identify cluster centers in low-density regions. Moreover, the single-step allocation strategy of DPC algorithm lacks fault tolerance. If a sample point is incorrectly allocated, it can trigger a cascade of subsequent allocation error. To solve the aforementioned issue, a density peak clustering algorithm based on natural neighbors and diffusion allocation (NNDA-DPC) is proposed. Firstly, the concept of natural neighbors is introduced, and the local density of sample points is redefined by considering the density of sample points and the environment in which the sample points are located. This is done to reduce the influence of the sparsity of class clusters on the selection of class cluster centers. The clustering centers are automatically determined by computing the decision value. Secondly, the natural neighbor points of the selected clustering centers are assigned to the corresponding clustering centers to form the initial microclusters. Finally, the initial microclusters are formed based on the remaining sample points and the initial clusters. The final clustering result is obtained by assigning the sample points based on the correlation between the remaining sample points and the initial clusters, which helps prevent the cascading effect of assignment errors. We compared the NNDA-DPC algorithm with DPC and its enhanced versions on artificial and UCI datasets. The experimental results demonstrate that the NNDA-DPC algorithm can efficiently cluster the datasets.

Use of a fuzzy granulation–degranulation criterion for assessing cluster validity

An image segmentation method based on a modified local-information weighted intuitionistic Fuzzy C-means clustering and Gold-panning Algorithm

Active contour model driven by Self Organizing Maps for image segmentation

Partitioning clustering has been one of the key components of data analytics to discover meaningful patterns in agricultural big data, driven by the increasing use of IoT-based technologies in smart farming. In partitioning clustering, the quality of clustering or performances of clustering algorithms are mostly evaluated by using the internal validity indices. In this study, the effectiveness of some widely used internal fuzzy indices are compared using the basic Fuzzy C-Means clustering algorithm. It is especially aimed to investigate changes in the effectiveness of validity indices when fuzzy data points are at different distances from the cluster centers. According to the results obtained on the simulated two-dimensional datasets, Fuzzy Silhouette, Fuzzy Hypervolume and Kwon are the most successful indices in validation of fuzzy clustering results.

The well known fuzzy partition clustering algorithms are most based on Euclidean distance function, which can only be used to detect spherical structural clusters. Gustafson-Kessel (GK) clustering algorithm and Gath-Geva (GG) clustering algorithm, were developed to detect non-spherical structural clusters, but both of them based on semi-supervised Mahalanobis distance needed additional prior information. An improved fuzzy C-mean algorithm based on unsupervised Mahalanobis distance, FCM-M, was proposed by our previous work, but it didn't consider the relationships between cluster centers in the objective function. In this paper, we proposed an improved fuzzy C-mean algorithm, FCM-MS, which is not only based on unsupervised Mahalanobis distance, but also considering the relationships between cluster centers, and the relationships between the center of all points and the cluster centers in the objective function, the singular and the initial values problems were also solved. A real data set was applied to prove that the performance of the FCM-MS algorithm gave more accurate clustering results than the FCM and FCM-M methods, and the ratio method which is proposed by us is the better of the two methods for selecting the initial values.

In a competitive market, it is of great significance to divide customer groups to develop customer-centered personalized products. In this paper, we propose a customer segmentation method based on the K-means algorithm and the improved particle swarm optimization (PSO) algorithm. As the PSO algorithm easily falls into local extremum, the improved hybrid particle swarm optimization (IHPSO) algorithm is proposed to improve optimization accuracy. The full factorial design is used to determine the optimal parameter combination; the roulette operator is used to select excellent particles; then, the selected particles are crossed according to their adaptive crossover probabilities; when the population falls into a local optimum, the particles are mutated according to their adaptive mutation probabilities. Aimed at the K-means’ sensitivity to selecting the initial cluster centers, IHPSO is used to optimize the cluster centers (IHPSO-KM). We compare IHPSO with the PSO, LDWPSO, GA, GA-PSO and ALPSO algorithms on nine benchmark functions. We also conduct comparative experiments to compare IHPSO-KM with several conventional and state-of-the-art approaches on five UCI datasets. All results show that the two proposed methods outperform existing models. Finally, IHPSO-KM is applied in customer segmentation. The experimental results also prove the rationality and applicability of IHPSO-KM for customer segmentation.

Recently, clustering algorithms combined conventional methods and artificial intelligence. FSC-SOM is designed to handle the problem of SOM, such as defining the number of clusters and initial value of neuron weights. FSC find the number of clusters and the cluster centers which become the parameter of SOM. FSC-SOM is expected to improve the quality of FSC since the determination of the cluster centers are processed twice i.e. searching for data with high density at FSC then updating the cluster centers at SOM. FSC-SOM was tested using 10 datasets that is measured with F-Measure, entropy, Silhouette Index, and Dunn Index. The result showed that FSC-SOM can improve the cluster center of FSC with SOM in order to obtain the better quality of clustering results. The clustering result of FSC-SOM is better than or equal to the clustering result of FSC that proven by the value of external and internal validity measurement.

<!--[if gte mso 10]> <mce:style><! /* Style Definitions */ table.MsoNormalTable {mso-style-name:"Norm&aacute;l t&aacute;bl&aacute;zat"; mso-tstyle-rowband-size:0; mso-tstyle-colband-size:0; mso-style-noshow:yes; mso-style-priority:99; mso-style-qformat:yes; mso-style-parent:""; mso-padding-alt:0cm 5.4pt 0cm 5.4pt; mso-para-margin:0cm; mso-para-margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:11.0pt; font-family:"Calibri","sans-serif"; mso-ascii-font-family:Calibri; mso-ascii-theme-font:minor-latin; mso-fareast-font-family:"Times New Roman"; mso-fareast-theme-font:minor-fareast; mso-hansi-font-family:Calibri; mso-hansi-theme-font:minor-latin; mso-bidi-font-family:"Times New Roman"; mso-bidi-theme-font:minor-bidi;} > <! [endif] ><p class="MsoNormal" style="text-align: justify; line-height: 200%;" _mce_style="text-align: justify; line-height: 200%;"><span style="font-size: 10pt; line-height: 200%; font-family: ";Arial";,";sans-serif";;" lang="EN-US">Traditional techniques of identification of a depositional body from core data are costly and sometimes difficult to extrapolate to uncored wells. However, application of Kohonen&rsquo;s Self Organized Map (SOM) approach may be regarded to be a potential method for pattern recognition problems. A combination of Indicator Kriging and SOM for log-porosity and sand content data coming from quantitative well-log interpretations is used for identifying the spatial pattern of some delta-plain sub-environments. The basic high-dimensional property fields are defined by 3D shapes of well known depositional facieses. Many parameters as log-porosity and sand content data can be used to determine geo-property as a lithological pattern using SOM. This step of method can discover spatial patterns as clusters in unstructured data set because SOM is based on clustering algorithm. However, this approach not necessarily makes sure, that the resulting disjunctive clusters can show any meaningful depositional geometry. So at last the final geometry is given using Indicator Kriging method, which uses threshold values derived from property values of clusters.</span><-->Traditional techniques to identify a depositional body from core data are costly and sometimes difficult to extrapolate to uncored wells. The application of Kohonen&rsquo;s Self Organized Map (SOM) approach may be useful for the interpretation of a depositional rock body through well-log data. SOM is based on a clustering algorithm and this method can be used to discover spatial patterns occurring as clusters in unstructured data sets. An example of the application of SOM is presented whereby clusters through SOM can indicate the contours of well-known depositional patterns such as sub-environments.

Fast global k-means clustering based on local geometrical information