Self-organizing Map Clustering Algorithm Research Articles

Lower tropospheric states revealed in high vertical-resolution radiosonde data in Korea and synoptic patterns for instability based on a self-organizing map

Seasonal and spatial characterizations of wind and stability in the lower troposphere below 1500 m are investigated using 1 s high-resolution operational radiosonde data observed at five stations (Baengnyeongdo, Heuksando, National Typhoon Center (NTC), Bukgangneung, and Pohang) in South Korea for 4 years (July 2016–June 2020), concentrating on four variables: horizontal wind speed (HWS), vertical wind shear (VWS), squared Brunt–Väisälä frequency (N2), and Richardson number (Ri). Significant seasonal variations are identified in these variables due to the heterogeneity of the geographical locations as well as seasonal variations in the synoptic patterns. VWS is high in spring and summer at two island stations on the west of the Korean Peninsula, Baengnyeongdo, and Heuksando, whereas it is high in winter at NTC, Bukgangneung, and Pohang, leading to frequent Kelvin–Helmholtz (KH) instability (KHI; 0 ≤ Ri < 1/4). Below z = 1000 m, static stability is lower in winter than in summer at all stations, resulting in a higher convective instability (Ri < 0) in winter, which is attributed to the inflow of the marine boundary layer. The frequent convective instability observed in spring at Bukgangneung suggests a potential relationship with mountain wave breaking and downslope windstorms. The microstructures of the lower tropospheric instabilities revealed in the current radiosonde observations are not properly represented in the high-resolution local numerical weather prediction and analysis system, namely local Data Assimilation and Prediction System (LDAPS) of Korea Meteorological Administration. In particular, the observed convective and KH instabilities, which are approximately 45% for the whole period, are rarely (approximately 15%) represented in LDAPS. The synoptic patterns favorable for the convective and KH instabilities cases are identified based on a self-organizing map clustering algorithm: (1) west-high and east-low patterns with extended Siberian high in winter, (2) strong low-pressure systems susceptible to KHI from spring to fall, (3) weak high-pressure systems with potential convective instability, accompanying northerly or north-westerly flows with cold advection at 925 hPa.

Stochastic green profit-maximizing hub location problem

This article proposes a two-stage stochastic profit-maximizing hub location problem (HLP) with uncertain demand. Additionally, the model incorporates several carbon regulations, such as carbon tax policy (CTP), carbon cap-and-trade policy (CCTP), carbon cap policy (CCP), and carbon offset policy (COP). In the proposed models, an enhanced sample average approximation (ESAA) method was used to obtain a suitable number of scenarios. To cluster similar samples, k-means clustering and self-organizing map (SOM) clustering algorithms were embedded in the ESAA. The L-shaped algorithm was employed to solve the model inside the ESAA method more efficiently. The proposed models were analyzed using the well-known Australian Post (AP) data set. Computational experiments showed that all of the carbon regulations could reduce overall carbon emissions. Among carbon policies, CCTP could achieve better economic results for the transportation sector. The results also demonstrated that the SOM clustering algorithm within the ESAA method was superior to both k-means inside ESAA and classical SAA algorithms according to the %gap and standard deviation measures. In addition, the results showed that the L-shaped algorithm performed better than the commercial solver in large-scale instances.

Urban Flood Risk Assessment in Zhengzhou, China, Based on a D-Number-Improved Analytic Hierarchy Process and a Self-Organizing Map Algorithm

Flood risk assessment is an important tool for disaster warning and prevention. In this study, an integrated approach based on a D-number-improved analytic hierarchy process (D-AHP) and a self-organizing map (SOM) clustering algorithm are proposed for urban flooding risk assessment. The urban flood inundation model and geographic information system (GIS) technology were used to quantify the assessment indices of urban flood risk. The D-AHP approach was adopted to determine the weights of the indices, which effectively makes up for the shortcomings of the AHP in dealing with uncertain evaluation information (such as fuzzy and incomplete information). In addition, the SOM clustering algorithm was applied to determine the flood risk level. It is a data-driven approach that avoids the subjective determination of a flood risk classification threshold. The proposed approach for flood risk assessment was implemented in Zhengzhou, China. The flood risk was classified into five levels: highest risk, higher risk, medium risk, lower risk, and the lowest risk. The proportion of the highest risk areas was 9.86%; such areas were mainly distributed in the central and eastern parts of the Jinshui District, the eastern part of the Huiji District, and the northeastern part of the Guancheng District, where there were low terrain and serious waterlogging. The higher risk areas accounted for 24.26% of the study area, and were mainly distributed in the western and southern parts of the Jinshui District, the southern part of the Huiji District, the middle and eastern parts of the Zhongyuan District, the northeastern part of the Erqi District, and the northwestern part of the Guancheng District, which consisted of economically developed areas of dense population and buildings, matching well with historical flooding events. To verify the effectiveness of the proposed approach, traditional approaches for risk assessment were compared. The comparison indicated that the proposed approach is more reasonable and accurate than the traditional approaches. This study showed the potential of a novel approach to flood risk assessment. The results can provide a reference for urban flood management and disaster reduction in the study area.

An improved Kohonen self-organizing map clustering algorithm for high-dimensional data sets

&lt;p&gt;Manipulating high-dimensional data is a major research challenge in the field of computer science in recent years. To classify this data, a lot of clustering algorithms have already been proposed. Kohonen self-organizing map (KSOM) is one of them. However, this algorithm has some drawbacks like overlapping clusters and non-linear separability problems. Therefore, in this paper, we propose an improved KSOM (I-KSOM) to reduce the problems that measures distances among objects using EISEN Cosine correlation formula. So far as we know, no previous work has used EISEN Cosine correlation distance measurements to classify high-dimensional data sets. To the robustness of the proposed KSOM, we carry out the experiments on several popular datasets like Iris, Seeds, Glass, Vertebral column, and Wisconsin breast cancer data sets. Our proposed algorithm shows better result compared to the existing original KSOM and another modified KSOM in terms of predictive performance with topographic and quantization error.&lt;/p&gt;

A multi-stage method to predict carbon dioxide emissions using dimensionality reduction, clustering, and machine learning techniques

The main purpose of this paper is to develop an efficient multi-stage methodology to predict carbon dioxide emissions based on two important variables including the energy consumption and economic growth using the clustering, prediction machine learning techniques, and dimensionality reduction. To do so, we use the self-organizing map clustering algorithm to cluster the data and the adaptive neuro-fuzzy inference system and artificial neural network to construct the prediction models in each cluster of the self-organizing map to predict carbon dioxide emissions considering a set of input parameters including economic growth and energy consumption in Group 20 nations. Furthermore, we use the singular value decomposition for dimensionality reduction and missing values’ prediction in the dataset. The results of the analysis of a real-world dataset found that the developed multi-stage approach was capable of predicting the carbon dioxide emissions on two indicators. To validate the proposed method, the results are compared with other existing methods. The outcomes demonstrate that the adaptive neuro-fuzzy inference system and artificial neural network techniques combined with the self-organizing map and singular value decomposition technique provide 0.065 accuracy in terms of the mean average error. In addition, when comparing singular value decomposition-self-organizing map-adaptive neuro-fuzzy inference system method with the singular value decomposition-self-organizing map-adaptive-artificial neural network method, the singular value decomposition-self-organizing map-adaptive neuro-fuzzy inference provides with 0.104 accuracy in predicting CO2 emissions. Moreover, the multiple linear regression provides the worst accuracy (0.522) results compared with the artificial neural network and adaptive neuro-fuzzy inference system techniques. The analysis regarding the relationship between economic development, carbon dioxide emissions, and the energy consumption is extremely vital from the energy and economic policy-making aspects in Group 20 countries given that the primary focus of this group has been the governance of the global economy.

Uncertainty Modeling of a Group Tourism Recommendation System Based on Pearson Similarity Criteria, Bayesian Network and Self-Organizing Map Clustering Algorithm

Uncertainty Modeling of a Group Tourism Recommendation System Based on Pearson Similarity Criteria, Bayesian Network and Self-Organizing Map Clustering Algorithm

Multi-Objective Optimized Aggregation of Demand Side Resources Based on a Self-organizing Map Clustering Algorithm Considering a Multi-Scenario Technique

To fully investigate the characteristics and the complementarities of demand side resources (DSRs), and to achieve efficient utilization of resources, the aggregation of DSRs is studied in this paper. Considering the uncertainty of DSRs, the characteristics analysis and the selection of relevant daily feature corresponding to various types of DSR are carried out. Then a multi-scenario model based on quarter division and self-organizing map (SOM) neural network algorithm is proposed. In the model, the clustering feature vector is selected as the input vector of the SOM algorithm to perform DSR clustering analysis to get the different scenarios. In addition, to obtain the resource aggregation (RA) with good load characteristics, response characteristics and distributed generation (DG) consumption, a multi-scenario objective optimization aggregation model of DSR based on scenario partition is established, and an the model is solved by an improved niche evolutionary multi-objective immune algorithm. Finally, the case studies are given to verify the validity of the model.

Load estimation for microgrid planning based on a self-organizing map methodology

This study presents a novel load estimation method for isolated communities that do not receive energy or only receive it for a limited time each day. These profiles have been used to determine the installed capacity of generating units for microgrid electrification projects. The social characteristics and lifestyles of isolated communities differ from those in urban areas; therefore, the load profiles of microgrids are sensitive to minor variations in generation and/or consumption. The proposed methodology for obtaining the residential profiles is based on clustering algorithms such as k-means, a self-organizing map (SOM) or others. In this work, SOM clustering is considered because it allows a better interpretation of results that can be contrasted with social aspects. The proposed methodology includes the following components. First, the inputs are processed based on surveys of residents that live in each socio-economic level of housing and the community. Second, family types are clustered using an SOM, from which relevant information is derived that distinguishes one family from another. Third, the load profiles of each cluster are selected from a database. Additionally, social aspects and relevant energy supply information from communities with similar characteristics are used to generate the required database. The SOM for the clustering of families of the community with available energy measurements is used as an initial guess for the clustering of the families in the community with unknown energy measurements.The methodology is applied and tested in the community of El Romeral, Chile, where a microgrid will be installed. The SOM technique compares favorably with a benchmark method that uses the average load profile of a community; furthermore, the SOM clustering algorithm for the methodology is favorably compared with the k-means algorithm because the results obtained by SOM are consistent with the social aspects.

A novel stochastic modeling method of wind power time series considering the fluctuation process characteristics

With rapid increasing of wind power generating capacity, the influence of wind power on power system planning and operation becomes more and more severe. Wind power time series modeling is the basis of evaluating the influence accurately of wind power on power system, which is very important to annual wind power dispatch planning and installed capacity optimization. In this paper, a novel modeling method of wind power time series based on the definition and identification of the fluctuation process is proposed. The Gaussian function is proved to be a good choice to sketch the fluctuation process and the visualization self-organizing map clustering algorithm is used to classify the fluctuation process. The expected wind power time series is achieved by studying the statistical feature of the sample data and by analyzing the transfer characteristic of the fluctuation processes. The comparative analysis on the statistical feature of the generated time series and the sample data is presented. In terms of the 15 min and 1 h variation, autocorrelation function, monthly mean and hourly mean of the generated wind power, the proposed method in this paper are better than the commonly used Markov Chain Monte Carlo method. The proposed method can keep the variation and process characteristic of the wind power, which is very important for the study of stochastic planning and operation of the wind power in power system.

A distributed decision framework for building clusters with different heterogeneity settings

In the past few decades, extensive research has been conducted to develop operation and control strategy for smart buildings with the purpose of reducing energy consumption. Besides studying on single building, it is envisioned that the next generation buildings can freely connect with one another to share energy and exchange information in the context of smart grid. It was demonstrated that a network of connected buildings (aka building clusters) can significantly reduce primary energy consumption, improve environmental sustainability and building’s resilience capability. However, an analytic tool to determine which type of buildings should form a cluster and what is the impact of building clusters’ heterogeneity based on energy profile to the energy performance of building clusters is missing. To bridge these research gaps, we propose a self-organizing map clustering algorithm to divide multiple buildings to different clusters based on their energy profiles, and a homogeneity index to evaluate the heterogeneity of different building clusters configurations. In addition, a bi-level distributed decision model is developed to study the energy sharing in the building clusters. To demonstrate the effectiveness of the proposed clustering algorithm and decision model, we employ a dataset including monthly energy consumption data for 30 buildings where the data is collected every 15min. It is demonstrated that the proposed decision model can achieve at least 13% cost savings for building clusters. The results show that the heterogeneity of energy profile is an important factor to select battery and renewable energy source for building clusters, and the shared battery and renewable energy are preferred for more heterogeneous building clusters.

Elementary Flux Mode Analysis Revealed Cyclization Pathway as a Powerful Way for NADPH Regeneration of Central Carbon Metabolism.

NADPH regeneration capacity is attracting growing research attention due to its important role in resisting oxidative stress. Besides, NADPH availability has been regarded as a limiting factor in production of industrially valuable compounds. The central carbon metabolism carries the carbon skeleton flux supporting the operation of NADPH-regenerating enzyme and offers flexibility in coping with NADPH demand for varied intracellular environment. To acquire an insightful understanding of its NADPH regeneration capacity, the elementary mode method was employed to compute all elementary flux modes (EFMs) of a network representative of central carbon metabolism. Based on the metabolic flux distributions of these modes, a cluster analysis of EFMs with high NADPH regeneration rate was conducted using the self-organizing map clustering algorithm. The clustering results were used to study the relationship between the flux of total NADPH regeneration and the flux in each NADPH producing enzyme. The results identified several reaction combinations supporting high NADPH regeneration, which are proven to be feasible in cells via thermodynamic analysis and coincident with a great deal of previous experimental report. Meanwhile, the reaction combinations showed some common characteristics: there were one or two decarboxylation oxidation reactions in the combinations that produced NADPH and the combination constitution included certain gluconeogenesis pathways. These findings suggested cyclization pathways as a powerful way for NADPH regeneration capacity of bacterial central carbon metabolism.

Visualizing text similarities from a graph-based SOM

Text in articles is based on expert opinion of a large number of people including the views of authors. These views are based on cultural or community aspects, which make extracting information from text very difficult. This paper introduced how to utilize the capabilities of a modified graph-based Self-Organizing Map (SOM) in showing text similarities. Text similarities are extracted from an article using Google's PageRank algorithm. Sentences from an input article are represented as graph model instead of vector space model. The resulted graph can be shown in a visual animation for eight famous graph algorithms execution with animation speed control.The resulted graph is used as an input to SOM. SOM clustering algorithm is used to construct knowledge from text data. We used a visual animation for eight famous graph methods with animation speed control and according to similarity measure; an adjustable number of most similar sentences are arranged in visual form. In addition, this paper presents a wide variety of text searching. We had compared our project with famous clustering and visualization project in term of purity, entropy and F measure. Our project showed accepted results and mostly superiority over other projects.

An Improved Correlation Measure-based SOM Clustering Algorithm for Gene Selection

Among the large amount of genes presented in microarray gene expression data, only a small fraction of them is effective for performing a certain diagnostic test. For this reason, reducing the dimensionality of gene expression data is imperative. Self-organizing map (SOM) is a type of mathematical cluster analysis which particularly well suited for recognizing and classifying features in complex, multidimensional data. This paper proposes an improved Self-organizing map clustering algorithm which based on neighborhood mutual information correlation measure. To evaluate the performance of the proposed approach, we apply it to six well-known gene expression datasets and compare our results with those obtained by other methods. Finally, the experimental results show that the proposed approach to gene selection is indeed efficient.

Artificial neural networks as potential classification tools for dinoflagellate cyst images: A case using the self-organizing map clustering algorithm

Automated palynological analysis has been previously proposed but proved difficult to achieve. Here, the first instance of dinoflagellate cyst (dinocyst) image clustering from palynological samples based on morphological and textural image analysis (IA) features is presented. Dinocyst-dominated example images (including acritarchs and other algae) were acquired from Cretaceous, Paleogene and Holocene samples. IA features that cover a broad array of measurements are used, including morphological, Fourier and textural descriptors, as well as geometric moments and color. To determine clusters, unsupervised self- organized maps (a genre of artificial neural networks) were used. Self-organized maps (SOMs) can determine their own classification of data in cases where no knowledge of the true class of each data point exists. Using a SOM five major clusters can be identified including clusters of freshwater alga, proximate dinocysts, proximate dinocysts with long horns, proximochorate dinocysts and chorate dinocysts. Minor variations can also be identified based on red, green and blue color, textural variations, and dinocyst process length in the form of morphological (shape) descriptors. These major and minor clusters demonstrate the open-ended capability of the system to be a wider palynological identification tool. The advantages of automation in palynology are articulated and the hurdles yet to overcome discussed. These advantages include freeing humans from ‘routine identification’ so that more emphasis can be placed upon distinguishing the rarer (‘unknown’) particles and placing them in a descriptive context. Challenges yet to conquer include distinguishing between images of closely related morphologies and/or textures. It is envisaged that technological and computational developments will quickly facilitate such further developments.

Phenotypic Anchoring of Global Gene Expression Profiles Induced by N-Hydroxy-4-acetylaminobiphenyl and Benzo[a]pyrene Diol Epoxide Reveals Correlations between Expression Profiles and Mechanism of Toxicity

The goal of this study was to compare changes in gene expression induced by exposure to different carcinogens and to anchor these changes to the induced levels of toxicity and mutagenesis. The human TK6 lymphoblastoid cell line was used as an in vitro model system, and reactive metabolites of two human carcinogens, benzo[a]pyrene and 4-aminobiphenyl, were used as model compounds. We first determined the toxicity of the model compounds N-hydroxy-4-acetylaminobiphenyl (N-OH-AABP) and benzo[a]pyrene diol epoxide (BPDE) in TK6 cells. BPDE was about 1000-fold more toxic and mutagenic than N-OH-AABP in TK6 cells on a molar basis. We next treated cells with three doses of each compound that resulted in low, medium, and high toxicities (5, 15, and 40%) and harvested cells at different times after exposure. Using comparable levels of toxicity as the phenotypic anchor, we compared the patterns of gene expression induced by each reactive metabolite using printed cDNA microarrays comprising approximately 18,000 human gene/EST sequences. The microarray data from the N-OH-AABP and BPDE treatment groups were compared using self-organizing map clustering algorithms, as well as a statistical regression modeling approach. While subsets of genes indicative of a generalized stress response [Hsp 40 homologue (DNAJ), Hsp70, Hsp105, and Hsp 125] were detected after exposure to both compounds at all concentrations, there were also many differentially regulated genes, including phase I xenobiotic metabolism [e.g., glutathione transferase omega (GSTTLp28) and antioxidant enzymes (Apxl)]. Other differentially regulated genes included those encoding proteins involved in all major DNA repair pathways, including excision repair (e.g., ERCC5), mismatch repair (e.g., MLH3), damage specific DNA binding protein (e.g., DDB2), and cisplatin resistance-associated overexpressed protein (LUC7A, CRA). Differences in the transcriptional response of TK6 cells to N-OH-AABP or BPDE exposure may explain the dramatic differences in the toxicity and mutagenicity of these human carcinogens.

Hybrid mining approach in the design of credit scoring models

Unrepresentative data samples are likely to reduce the utility of data classifiers in practical application. This study presents a hybrid mining approach in the design of an effective credit scoring model, based on clustering and neural network techniques. We used clustering techniques to preprocess the input samples with the objective of indicating unrepresentative samples into isolated and inconsistent clusters, and used neural networks to construct the credit scoring model. The clustering stage involved a class-wise classification process. A self-organizing map clustering algorithm was used to automatically determine the number of clusters and the starting points of each cluster. Then, the K-means clustering algorithm was used to generate clusters of samples belonging to new classes and eliminate the unrepresentative samples from each class. In the neural network stage, samples with new class labels were used in the design of the credit scoring model. The proposed method demonstrates by two real world credit data sets that the hybrid mining approach can be used to build effective credit scoring models.