The self-organizing map

2009

Dynamic environments represent a quite complex domain, where the information available changes continuously. In this paper, a contingency response system for dynamic environments called MACSDE is presented. The explained system allows the introduction of information, the monitoring of the process and the generation of predictions. The system makes use of a Case-Based Reasoning system which generates predictions using previously gathered information. It employs a distributed multi-agent architecture so that the main components of the system can be accessed remotely. Therefore, all functionalities can communicate in a distributed way, even from mobile devices. The core of the system is a group of deliberative agents acting as controllers and administrators for all functionalities. The system explained includes a novel network for data classification and retrieval. Such network works as a summarization algorithm for the results of an ensemble of Self-Organizing Maps. The presented system has been tested with data related with oil spills and forest fire, obtaining quite hopeful results.

Forest fires represent a quite complex environment and an accurate prediction of the fires generated is crucial when trying to react quickly and effectively in such a critical situation. In this study, an hybrid system is applied to predict the evolution of forest fires. The Case-Based Reasoning methodology combined with a summarization of SOM ensembles algorithm has been used to face this problem. The CBR methodology is used as the solution generator in the system, reusing past solutions given to past problems to generate new solutions to new problems by adapting those past solutions to the new situations to face. On the other hand, a new summarization algorithm (WeVoS-SOM) is used to organize the stored information to make it easier to retrieve the most useful information from the case base. The developed system has been checked with forest fires historical and experimental data. The WeVoS-CBR system presented here has successfully predicted the evolution of the forest fires in terms of probability of finding fires in a certain area.

International Journal of Advanced Manufacturing Technology, 2011

Machine–part cell formation is used in cellular manufacturing in order to process large varieties, improve quality, and lower work-in-process levels, reducing manufacturing lead time and customer response time while retaining flexibility for new products. This paper presents a new and novel approach for obtaining machine cells and part families. In cellular manufacturing, the fundamental problem is the formation of part families and machine cells. The present paper deals with the self-organizing map (SOM) method, an unsupervised learning algorithm in artificial intelligence which has been used as a visually decipherable clustering tool of machine–part cell formation. The objective of the paper is to cluster the binary machine–part matrix through visually decipherable cluster of SOM color coding and labeling via the SOM map nodes in such a way that the part families are processed in that machine cell. The U-matrix, component plane, principal component projection, scatter plot, and histogram of SOM have been reported in the present work for the successful visualization of the machine–part cell formation. Computational result with the proposed algorithm on a set of group technology problems available in the literature is also presented. The proposed SOM approach produced solutions with a grouping efficacy that is at least as good as any results earlier reported in the literature and improved the grouping efficacy for 70% of the problems and was found to be immensely useful to both industry practitioners and researchers.

2000

This study presents a novel hybrid intelligent system which focuses on the optimisation of machine parameters for dental milling purposes based on the following phases. Firstly, an unsupervised neural model extracts the internal structure of a data set describing the model and also the relevant features of the data set which represents the system. Secondly, the dynamic system performance of different variables is specifically modelled using a supervised neural model and identification techniques from relevant features of the data set. This model constitutes the goal function of the production process. Finally, a genetic algorithm is used to optimise the machine parameters from a non parametric fitness function. The reliability of the proposed novel hybrid system is validated with a real industrial use case, based on the optimisation of a high-precision machining centre with five axes for dental milling purposes.

Artificial Life, 2009

The current work addresses the development of brain-inspired models that will be embedded in robotic systems to support their cognitive abilities. We introduce a novel agent-based coevolutionary computational framework for modelling assemblies of brain areas. Specifically, self-organized agent structures are employed to represent brain areas. In order to support the design of agents, we introduce a Hierarchical Cooperative CoEvolutionary (HCCE) scheme that effectively specifies the structural details of autonomous, yet cooperating system components. The design process is facilitated by the capability of the HCCE-based design mechanism to investigate the performance of the model in lesion conditions. Interestingly enough, HCCE provides also a consistent mechanism to reconfigure (if necessary) the structure of agents, facilitating follow-up modelling efforts. Implemented models are embedded in a simulated robot to support its behavioral capabilities, demonstrating also the validity of the proposed computational framework.

Environmetrics, 2011

Fluorescence spectroscopy enables fast and sensitive analysis of environmental samples containing various organic matter constituents. However, to retrieve valuable information from fluorescence spectra, robust techniques for data analysis should be employed. Here, different multivariate analysis methods and artificial neural networks (ANNs) were applied for decomposition and calibration of fluorescence excitation-emission matrices (EEMs). This is the first paper summarizing the application of different data mining methods, from multiway analysis to ANNs, for fluorescence EEMs technique employed to characterize organic matter properties and removal in the field of drinking water treatment. Fluorescence analysis was carried out on municipal water treatment samples of raw and partially-treated water. Parallel factor analysis (PARAFAC) method and self-organizing maps were used to analyse EEMs, extract information on the organic matter constituents and reduce the dimensionality of the data to enhance the efficiency of calibration methods. Partial least squares (PLS), multiple linear regression (MLR) and neural network with back-propagation were employed for calibration of fluorescence data with actual total organic carbon (TOC) concentrations. All models except PARAFAC-MLR produced consistent results with correlation coefficient R 2 ¼ 0.93 for validation dataset. This is the first such comparative analysis of fluorescence data modelling that clarifies fundamental fluorescence data analysis questions regarding the suitability of different decomposition and calibration methods.

PLOS ONE, 2015

For infants, the first problem in learning a word is to map the word to its referent; a second problem is to remember that mapping when the word and/or referent are again encountered. Recent infant studies suggest that spatial location plays a key role in how infants solve both problems. Here we provide a new theoretical model and new empirical evidence on how the bodyand its momentary posturemay be central to these processes. The present study uses a name-object mapping task in which names are either encountered in the absence of their target (experiments 1-3, 6 & 7), or when their target is present but in a location previously associated with a foil (experiments 4, 5, 8 & 9). A humanoid robot model (experiments 1-5) is used to instantiate and test the hypothesis that body-centric spatial location, and thus the bodies' momentary posture, is used to centrally bind the multimodal features of heard names and visual objects. The robot model is shown to replicate existing infant data and then to generate novel predictions, which are tested in new infant studies (experiments 6-9). Despite spatial location being task-irrelevant in this second set of experiments, infants use body-centric spatial contingency over temporal contingency to map the name to object. Both infants and the robot remember the name-object mapping even in new spatial locations. However, the robot model shows how this memory can emerge -not from separating bodily information from the word-object mapping as proposed in previous models of the role of space in word-object mappingbut through the body's momentary disposition in space.

Journal of Geophysical Research, 2009

1] This paper illustrates the main results of a statistical analysis performed on a data set obtained by integrating experimental observations collected during many oceanographic research projects on the northern Adriatic Sea (NAS). The observations cover the last 20 years and provide a robust base for the assessment of the current state and scales of variability for temperature, salinity, nutrients, dissolved oxygen, and chlorophyll. The results confirmed a clear seasonal cycle and marked spatial gradients for most parameters in all seasons. The largest proportion of the river Po input flows south along the coast, but significant eastward freshwater fluxes are also present in all seasons, more markedly in winter. The coastal belt south of the Lagoon of Venice is the most eutrophic area, mainly because of river inputs, while an oligotrophic condition prevails along the eastern part of the basin. Small-scale structures, including eddies and jets, are permanent features of the system. In order to test the existence of significant trends of variation in the physical and biogeochemical parameters, the data set has been enlarged by including observations from 1976. Analyses of trends over 30 years show an increase in salinity, which might be a consequence of both reduced outflows from rivers and a more sustained inflow of water along eastern coast, and a clear reduction in concentrations of phosphate and ammonia in coastal areas, probably due to new regulations regarding the control of nutrient loads and possibly suggesting the occurrence of cultural oligotrophication. No decrease is instead observed for concentration of nitrate.

Journal of Geophysical Research, 2007

1] Projections from 15 global climate models and 2 reanalysis products (National Center for Environmental Prediction (NCEP)/National Center for Atmospheric and Climate Research (NCAR) reanalysis (NNR) and European Centre for Medium-Range Weather Forecasts (ECMWF) 40-year reanalysis (ERA40)) were utilized to project changes in the net precipitation (P À E) over the Southern Ocean and Antarctica during the 21st century. Three time periods, 1979-2000, 2046-2055, and 2091-2100, of data were compared. The P À E was related to a classification of synoptic circulation patterns obtained using a neural network algorithm known as self-organizing maps (SOMs). SOM classification was successfully used as a quality control tool to assess the simulated atmospheric circulation and model performance in P À E. The models predicted an increase of Antarctic P À E that averages 0.42 ± 0.01 mm year À1 for the coming hundred years based on the difference between 1979-2000 and 2091-2100. P À E changes of individual models ranged from 0.02 to 0.71 mm year À1 . P À E integrated over the entire Antarctic ice sheet was forecast to increase more quickly from the end of the twentieth century until 2046-2055 than from 2046-2055 until 2091-2100. Contributions to the predicted change in P À E were evaluated for both thermodynamic and dynamic processes. The projected change in Antarctic P À E was primarily due to thermodynamic changes rather than circulation changes. The dynamic component of P À E change, associated with the circulation, was important at subcontinental scales, especially over the coastal regions. The role of dynamic changes was maintained until the end of the 21st century. Intermodel variation in predicted P À E changes and differences between models and reanalyses in the twentieth-century simulations severely restrict the reliability of these projections and highlight the need for improved polar simulations in climate models.

2010

Journal of Autism and Developmental Disorders, 2014

Varied cluster analysis were applied to facial surface measurements from 62 prepubertal boys with essential autism to determine whether facial morphology constitutes viable biomarker for delineation of discrete Autism Spectrum Disorders (ASD) subgroups. Earlier study indicated utility of facial morphology for autism subgrouping (Aldridge et al. in Mol Autism 2(1):15, 2011). Geodesic distances between standardized facial landmarks were measured from three-dimensional stereo-photogrammetric images. Subjects were evaluated for autism-related symptoms, neurologic, cognitive, familial, and phenotypic variants. The most compact cluster is clinically characterized by severe ASD, significant cognitive impairment and language regression. This verifies utility of facially-based ASD subtypes and validates Aldridge et al.'s severe ASD subgroup, notwithstanding different techniques. It suggests that language regression may define a unique ASD subgroup with potential etiologic differences.

Online Information Review, 2011

Emerald Article: Improving self-organising information maps as navigational tools: a semantic approach Yi-ling Lin, Peter Brusilovsky, Daqing He

Italian Journal of Animal Science, 2010

The work examines the hypothesis that the behavioural disorders found in pigs under conditions of stress may also be attributable to inherent conditions of alteration in mood. In light of new evidence that links the biochemical characterization of human depression to a particular profile of fatty acids in platelets, in particular the Palmitic Acid, Linoleic and Arachidonic, the decision was made to investigate platelet fatty acids of different animal species (pig, cattle, cat, sheep), together with the same data found in literature for rats and guinea pigs. The results obtained from normal and depressed human subjects have made it possible to achieve a particular Artificial Neural Network called the Self Organizing Map (SOM). This network, which is also used in the classification of some species (pigs, cattle, cats and sheep), has been utilised to distribute and classify all the animals studied, in agreement with the fatty acid markers of depressive disorder and the degree of saturation of membrane lipids. In agreement with this approach, the pig is comparable to humans that present a clinical diagnosis of depression. A critical analysis of specific references indicates the existence of a wide range of similarities between human beings suffering from depression and pigs. All the results we obtained on platelets, together with bibliographic evidence make plausible, in our view, the hypothesis that the pig is an animal intrinsically prone to depression. This tendency, which is probably genetically predetermined, must be taken into account in studies on the welfare of this animal and could also serve as a good model for the study of antidepressant molecules for humans.

PLoS ONE, 2014

DNA microarrays and cell cycle synchronization experiments have made possible the study of the mechanisms of cell cycle regulation of Saccharomyces cerevisiae by simultaneously monitoring the expression levels of thousands of genes at specific time points. On the other hand, pattern recognition techniques can contribute to the analysis of such massive measurements, providing a model of gene expression level evolution through the cell cycle process. In this paper, we propose the use of one of such techniques -an unsupervised artificial neural network called a Self-Organizing Map (SOM)which has been successfully applied to processes involving very noisy signals, classifying and organizing them, and assisting in the discovery of behavior patterns without requiring prior knowledge about the process under analysis. As a test bed for the use of SOMs in finding possible relationships among genes and their possible contribution in some biological processes, we selected 282 S. cerevisiae genes that have been shown through biological experiments to have an activity during the cell cycle. The expression level of these genes was analyzed in five of the most cited time series DNA microarray databases used in the study of the cell cycle of this organism. With the use of SOM, it was possible to find clusters of genes with similar behavior in the five databases along two cell cycles. This result suggested that some of these genes might be biologically related or might have a regulatory relationship, as was corroborated by comparing some of the clusters obtained with SOMs against a previously reported regulatory network that was generated using biological knowledge, such as protein-protein interactions, gene expression levels, metabolism dynamics, promoter binding, and modification, regulation and transport of proteins. The methodology described in this paper could be applied to the study of gene relationships of other biological processes in different organisms.

Environmental Progress & Sustainable Energy, 2011

The application of mathematical tools can be necessary to provide an integrated analysis and interpretation of the abundant information that can be collected in air quality monitoring networks. This article develops a methodology based on the use of Self-Organizing Map (SOM) artificial neural networks for integrating data about multiple measured pollutants to group monitoring stations according to their similar air quality. The proposed method considers the subsequent geographical mapping of the clusters of stations observed with the SOM, which can make it possible to detect geographically different areas but that share similar air pollution problems. This methodology is illustrated with its application to a case study in which 517 stations of the Spanish air quality monitoring network were classified considering simultaneously their levels of regulated pollutants in 2005, highlighting some implications of data normalization in the process. In particular, the use of legal limit values to normalize the concentrations of pollutants proved to be especially advisable. Results obtained with the SOM-based methodology, when compared to classifications based directly on legislation, provided more useful classifications for further air quality management actions, and revealed that these types of tools can facilitate the design of air pollution reduction programs by discovering different areas with similar problems.

Soft Computing, 2010

In this paper, an approach using fuzzy logic techniques and self-organizing maps (SOM) is presented in order to manage conceptual aspects in document clusters and to reduce the training time. In order to measure the presence degree of a concept in a document, a concept frequency formula is introduced. This formula is based on new fuzzy formulas to calculate the polysemy degree of terms and the synonymy degree between terms. In this approach, new fuzzy improvements such as automatic choice of the topology, heuristic map initialization, a fuzzy similarity measure and a keywords extraction process are used. Some experiments have been carried out in order to compare the proposed system with classic SOM approaches by means of Reuters collection. The system performance has been measured in terms of F-measure and training time. The experimental results show that the proposed approach generates good results with less training time compared to classic SOM techniques.

Complexity

The detection of frauds in credit card transactions is a major topic in financial research, of profound economic implications. While this has hitherto been tackled through data analysis techniques, the resemblances between this and other problems, like the design of recommendation systems and of diagnostic/prognostic medical tools, suggest that a complex network approach may yield important benefits. In this paper we present a first hybrid data mining/complex network classification algorithm, able to detect illegal instances in a real card transaction data set. It is based on a recently proposed network reconstruction algorithm that allows creating representations of the deviation of one instance from a reference group. We show how the inclusion of features extracted from the network data representation improves the score obtained by a standard, neural network-based classification algorithm and additionally how this combined approach can outperform a commercial fraud detection syste...

Quality & Quantity, 2013

In a service-dominant paradigm, service is the core of society and industry. This study explores the hierarchical roles of core service, facilitating service, and supporting service that affect customer satisfaction in the service industry. Cable TV services in Taiwan are employed as an example, because their penetration rate in is over 70% with customer experience for over 20 years. In this study, customers are first segmented by the self-organizing map (SOM) technique and then service quality is measured with SERVPREF method. A proposed model, derived from the service marketing literature, is tested with structural equation modeling (SEM). Finally, moderating effects of customer segments on the proposed SEM model are examined. Results show the SOM technique is a valid method for market segmentation. Additionally, the moderating effect of market segments may impact the evaluation of core service quality (CSQ) and overall satisfaction. Furthermore, CSQ determines overall satisfaction for heavy users, whereas facilitating service quality greatly influences younger consumers' satisfaction.

ISRN bioinformatics, 2014

Protein function prediction is a complex multiclass multilabel classification problem, characterized by multiple issues such as the incompleteness of the available annotations, the integration of multiple sources of high dimensional biomolecular data, the unbalance of several functional classes, and the difficulty of univocally determining negative examples. Moreover, the hierarchical relationships between functional classes that characterize both the Gene Ontology and FunCat taxonomies motivate the development of hierarchy-aware prediction methods that showed significantly better performances than hierarchical-unaware "flat" prediction methods. In this paper, we provide a comprehensive review of hierarchical methods for protein function prediction based on ensembles of learning machines. According to this general approach, a separate learning machine is trained to learn a specific functional term and then the resulting predictions are assembled in a "consensus" ...

Serials Review, 2009

Rating systems for journals often overlook the important issue of fit between journal and article. Strong fit is needed to obtain the most constructive review process which is critical to the eventual impact of the article. The relationship between journals is also important for decisions regarding the addition and cancellation of subscriptions from a collection of serials. We use a Kohonen selforganizing map as a visualization tool applied to business management literature to assess about 40,000 abstracts for 202 management journals listed in the Social Sciences Citation Index, the Science Citation Index, and the Financial Times list of business journals. We obtain a map which places journals with similar content very close together and journals with very different content far apart. This paper offers a method to consider how journals relate to each other and which journals are most and least likely to offer a fit with different types of research in the business management field.

SN Applied Sciences

This paper proposes the use of redundant features for efficient recognition of faces in still images using a novel system framework that offers a detailed systematic workflow for solving the facial recognition problem. It accepts still frontal face images, processes and represents salient facial features (face, eyes, nose, and mouth) using facial detection and extraction techniques. The extracted features are then modeled by an ensemble of self-organizing maps. The ensemble outputs are later reassembled into a single dataset consisting of a normalized image Euclidean distance matrix to enhance the search space for optimal convergence and classification. The feasibility of the framework is tested using an experiment facial database captured during the study, and three benchmark facial expression databases, namely the Extended Cohn-Kanade (CK+) database, the Japanese Female Facial Expressions database, and the MMI Facial Expression database. The results suggest that feature redundancy is indeed useful for efficient facial recognition, as the support vector machine classification recorded high accuracies across the various databases, with the normalized image Euclidean distance dataset producing the highest performance, when compared with the localized principal component analysis and unnormalized image Euclidean distance datasets. Furthermore, overall classification accuracy of above 99% was achieved for the experiment (nonexpressive still face) database, compared with the benchmark facial expression databases, which yielded slightly lower results. A future direction of this work is further improvement of the framework to robustly handle severe facial variations.

Frontiers in neurorobotics, 2018

The first "object" that newborn children play with is their own body. This activity allows them to autonomously form a sensorimotor map of their own body and a repertoire of actions supporting future cognitive and motor development. Here we propose the theoretical hypothesis, operationalized as a computational model, that this acquisition of body knowledge is not guided by random motor-babbling, but rather by autonomously generated goals formed on the basis of intrinsic motivations. Motor exploration leads the agent to discover and form representations of the possible sensory events it can cause with its own actions. When the agent realizes the possibility of improving the competence to re-activate those representations, it is intrinsically motivated to select and pursue them as goals. The model is based on four components: (1) a self-organizing neural network, modulated by competence-based intrinsic motivations, that acquires abstract representations of experienced sensor...

Scientific Reports

We aimed to evaluate the relationship between different lifestyle patterns and blood pressure. This study is based on the national survey of the risk factors for non-communicable diseases in Iran in 2012. A total of 8244 people aged 25-70 years old have been enrolled in the survey. Clustering on the individual data of lifestyle factors (nutrition, physical activity, and smoking) were carried out using self-organizing neural network method. Multivariable regression models were used to determine the relationship between blood pressure and the clusters. This study revealed seven lifestyle clusters in the national survey. The first cluster had a healthier lifestyle (15%), and the rest of the clusters had at least one or more lifestyle-related risk factors. Among all the clusters, people in two clusters, i.e. one characterized by consumption of sugar-sweetened beverages, salt, and fast foods, and the other one characterized by physical inactivity, were more exposed to the risk of hypertension (odds ratios of 1.44 and 1.12, respectively). People in another cluster who were 100% smokers and had a very high level of workrelated physical activity were about 30% less likely to experience elevated blood pressure. Although a lifestyle with cigarette smoking was associated with a reduction in blood pressure, this might be due to other related factors, such as work-related physical activity, which lower blood pressure. Of course, this hypothesis still needs to be further studied in the future. Elevated blood pressure is one of the most important health problems, which is considered as one of the major risk factors for cardiovascular disease, stroke, and kidney diseases as well 1,2. Various studies have shown that the prevalence of hypertension is increasing in various countries 3-5 , however, the highest level of increase has been observed in developing countries, especially in Asia and the East Mediterranean region. According to the Iranian Health Profile Survey, around 22% of the adult population are hypertensive and the problem is more prevalent in the population above 55 years old. The maximum and minimum rate of hypertension within Iran is in Tehran and Qazvin with the prevalence of 47% and 6.9%, respectively. In Iran, the prevalence of hypertension is higher in urban areas 6. According to many studies, this increase is due to changes in lifestyle and an increase in behavioral risk factors such as unhealthy nutrition, physical inactivity, and smoking 7-10. So far, many different studies have separately examined the relationship between different lifestyle-related factors and blood pressure, but these studies have neglected the fact that lifestyle-related risk factors usually occur together in a person and tend to be clustered. The clustering of health-related behavioral risk factors is a major threat to the public health. Recent studies have shown that the coincidence of health-related risk factors in a person can result in synergistic effects that in turn increase the risk of non-communicable diseases and subsequent mortality 11,12. A cohort study that was conducted on 20,244 Americans after 11 years of follow-up showed that people with all the four major risk factors to health (cigarette smoking, alcohol consumption, physical inactivity, and poor nutrition), compared with those

Recent advances, in high-throughput technologies allows whole transcriptome analysis, providing a complete and panoramic view of intragenic differential expression in eukaryotes. However, intragenic differential expression in prokaryotes still mystery and incompletely understood. In this study, we investigated and collected the evidence for intragenic differential expression in several archaeal transcriptomes such as, Halobacterium salinarum NRC-1, Pyrococcus furiosus, Methanococcus maripaludis, and Sulfolobus solfataricus, based on computational methods; specifically, by well-known self-organizing map (SOM) for cluster analysis, which transforms high dimensional data into low dimensional. We found 104 (3.86%) of genes in Halobacterium salinarum NRC-1, 59 (2.56%) of genes in Pyrococcus furiosus, 43 (2.41%) of genes Methanococcus maripaludis and 13 (0.42%) of genes in Sulfolobus solfataricus have two or more clusters, i.e., showed the intragenic differential expression at different c...

Science Signaling, 2015

Individual signaling pathways are not isolated, but rather operate in the context of the broader signaling network. Thus, the response of a cell to perturbation of a given pathway depends on the state of the network, which depends upon contextual inputs from the microenvironment. The cytokine tumor necrosis factor α (TNF-α) promotes opposing cellular behaviors under different conditions, which is influenced by perturbation of the network. For example, inhibition of the mitogen-activated protein kinase (MAPK) kinase MEK alters the kinetics of TNF-α-induced apoptosis in the mouse intestinal epithelium. We investigated whether MAPK signaling directly influences TNF-α-induced apoptosis, or whether network-level effects secondary to inhibition of the MAPK pathway alter the kinetics of cell death. We found that inhibitors of the MAPK kinase kinase Raf, MEK, and extracellular signal regulated kinase (ERK) exerted distinct effects on the timing and magnitude of TNF-α-induced apoptosis in the mouse intestine. Furthermore, even different MEK inhibitors exerted distinct effects; one of them, CH5126766, potentiated TNF-αinduced apoptosis. Computational modeling analysis and experimental perturbation identified the kinase Akt as the primary signaling node that promoted apoptosis in the context of TNF-α signaling in the presence of CH5126766. Our work emphasizes the importance of integrated network signaling in specifying cellular behavior in response to external perturbation. More †

PLOS ONE

Aquaculture is currently a major source of fish and has the potential to become a major source of protein in the future. These demands require efficient aquaculture. The intestinal microbiota plays an integral role that benefits the host, providing nutrition and modulating the immune system. Although our understanding of microbiota in fish gut has increased, comprehensive studies examining fish microbiota and host metabolism remain limited. Here, we investigated the microbiota and host metabolism in the coral leopard grouper, which is traded in Asian markets as a superior fish and has begun to be produced via aquaculture. We initially examined the structural changes of the gut microbiota using next-generation sequencing and found that the composition of microbiota changed between fasting and feeding conditions. The dominant phyla were Proteobacteria in fasting and Firmicutes in feeding; interchanging the dominant bacteria required 12 hours. Moreover, microbiota diversity was higher under feeding conditions than under fasting conditions. Multivariate analysis revealed that Proteobacteria are the key bacteria in fasting and Firmicutes and Fusobacteria are the key bacteria in feeding. Subsequently, we estimated microbiota functional capacity. Microbiota functional structure was relatively stable throughout the experiment; however, individual function activity changed according to feeding conditions. Taken together, these findings indicate that the gut microbiota could be a key factor to understanding fish feeding conditions and play a role in interactions with host metabolism. In addition, the composition of microbiota in ambient seawater directly affects the fish; therefore, it is important to monitor the microbiota in rearing tanks and seawater circulating systems.

PloS one, 2017

As the world's population grows to a projected 11.2 billion by 2100, the number of people living in low-lying areas exposed to coastal hazards is projected to increase. Critical infrastructure and valuable assets continue to be placed in vulnerable areas, and in recent years, millions of people have been displaced by natural hazards. Impacts from coastal hazards depend on the number of people, value of assets, and presence of critical resources in harm's way. Risks related to natural hazards are determined by a complex interaction between physical hazards, the vulnerability of a society or social-ecological system and its exposure to such hazards. Moreover, these risks are amplified by challenging socioeconomic dynamics, including poorly planned urban development, income inequality, and poverty. This study employs a combination of machine learning clustering techniques (Self Organizing Maps and K-Means) and a spatial index, to assess coastal risks in Latin America and the Ca...

Scientific Reports

Risks of stream fish homogenization are attributable to multiple variables operating at various spatial and temporal scales. However, understanding the mechanisms of homogenization requires not only watershed-scale, but also exhaustive fish community structure shifts representing detailed local functional relationships essential to homogenization potentials. Here, we demonstrate the idea of applying AI-based clusters to reveal nonlinear responses of homogenization risks among heterogeneous hydro-chemo-bio variables in space and time. Results found that species introduction, dam isolation, and the potential of climate-mediated disruptions in hydrologic cycles producing degradation in water quality triggered shifts of community assembly and resulting structures producing detrimental conditions for endemic fishes. The AI-based clustering approach suggests that endemic species conservation should focus on alleviation of low flows, control of species introduction, limiting generalist expansion, and enhancing the hydrological connectivity fragmented by dams. Likewise, it can be applied in other geographical and environmental settings for finding homogenization mitigation strategies. The homogenization process is often defined as the replacement of native species with a narrow geographic range by either native or non-indigenous species with a broad geographic range 1. The process generally increases diversity at a watershed scale when the endemic species co-occur with the native generalists or non-native species 2. Nonetheless, when extinction or extirpation of endemic species occurs 2,3 , the taxonomic diversity will be reduced at regional or global scales 4. One of the causes for homogenization has been documented as species expansion across their natural distribution boundaries 5 stimulated and/or accelerated intentionally or accidentally by human activities 6. Species expansion is mostly a result of species introduction, agriculture or aquaculture practices, and human transportation 7-9. Other documented biotic homogenization is triggered by, and often co-involved with, large-scale environmental alteration such as dam construction, channel habitat modification, and climate change 10,11. These modifications can, in some cases, produce expansion in the geographic range of some species, but may also lead to the extirpation of rare, localized, and endemic species 2,12. Contribution to species change can be found in the construction of shipping canals that promote the movement of formerly confined fish and invertebrate species across historical geographical constraints 13. Given that the local species composition is the result of interacting hydrologic, chemical, and biological (hydro-chemo-bio) factors, there should be a recognition that any existing ecological equilibria are due to simultaneous influence of a range of biotic and abiotic factors 14-16. Consequently, identification of homogenization is challenging because multiple dynamic process are involved. There may be subtle interactions among species, the species present both respond to, and reflect, the influence of environmental variables, and human-induced change may influence ecosystems over local to global scales. Extant ecosystems are complex with both structure and function influenced by species present, the interactions among species, and the response of each species to changes in local habitat and water quality conditions.

Applied Sciences

Circular economy promotes the use of waste materials into new production processes as a key factor for resource efficiency. The construction sector, and specifically the fired clay industry, is able to assimilate large amounts of waste in their processes, without significantly altering the technical properties of products. The introduction of different waste in ceramic products at the laboratory level has been extensively studied in the literature, but most of these studies have not yet been scaled-up to industrial production. Differences in processing with respect to laboratory conditions introduces uncertainty in relation to the expected properties of the final products. This paper uses a Self-Organizing Map (SOM)-based methodology for analysing and assessing the incorporation of industrial waste, Waelz slag (WS) and foundry sand dust (FSD), in ceramic products obtained sequentially at laboratory, semi-industrial and industrial level, over technological properties and metals relea...

Engineering Applications of Neural Networks

In this paper, a novel multiple-layer neuro-fuzzy network is proposed to model/predict shoreline realignment at a highly touristic island beach (Kamari beach, Santorini, Greece). A specialized experimental setup was deployed to generate a set of input-output data that comprise parameters describing the beach morphology and wave conditions and the cross-shore shoreline position at 30 cross-sections of the beach extracted from coastal video imagery, respectively. The proposed network consists of three distinct modules. The first module concerns the network representation of a fuzzy model equipped with a typical inference mechanism. The second module implements a novel competitive learning network to generate initial values for the rule base antecedent parameters. These parameters are, then, used to facilitate the third module that employs particle swarm optimization to perform a stochastic search for optimal parameter estimation. The network is compared favorably to two other neural networks: a radial basis function neural network and a feedforward neural network. Regarding the effectiveness of the proposed network to model shoreline realignment , the RMSE found (7.2-7.7 m, depending on the number of rules/nodes), reflects the high variability of the shoreline position of the Kamari beach during the period of observations: the RMSE is of a similar order to the standard deviation (up to 8 m) of the cross-shore shoreline position. The results are encouraging and the effectiveness of the proposed network could be further improved by changes (fine-tuning) of the input variables.

Identifying university students' weaknesses results in better learning and can function as an early warning system to enable students to improve. However, the satisfaction level of existing systems is not promising. New and dynamic hybrid systems are needed to imitate this mechanism. A hybrid system (a modified Recurrent Neural Network with an adapted Grey Wolf Optimizer) is used to forecast students' outcomes. This proposed system would improve instruction by the faculty and enhance the students' learning experiences. The results show that a modified recurrent neural network with an adapted Grey Wolf Optimizer has the best accuracy when compared with other models.

Proceedings of the 20th ACM International Conference on Multimodal Interaction

Automatic recognition of eating conditions of humans could be a useful technology in health monitoring. The audiovisual information can be used in automating this process, and feature engineering approaches can reduce the dimensionality of audiovisual information. The reduced dimensionality of data (particularly feature subset selection) can assist in designing a system for eating conditions recognition with lower power, cost, memory and computation resources than a system which is designed using full dimensions of data. This paper presents Active Feature Transformation (AFT) and Active Feature Selection (AFS) methods, and applies them to all three tasks of the ICMI 2018 EAT Challenge for recognition of user eating conditions using audio and visual features. The AFT method is used for the transformation of the Mel-frequency Cepstral Coefficient and ComParE features for the classification task, while the AFS method helps in selecting a feature subset. Transformation by Principal Component Analysis (PCA) is also used for comparison. We find feature subsets of audio features using the AFS method (422 for Food Type, 104 for Likability and 68 for Difficulty out of 988 features) which provide better results than the full feature set. Our results show that AFS outperforms PCA and AFT in terms of accuracy for the recognition of user eating conditions using audio features. The AFT of visual features (facial landmarks) provides less accurate results than the AFS and AFT sets of audio features. However, the weighted score fusion of all the feature set improves the results.

Geoscientific Instrumentation, Methods and Data Systems, 2017

The use of efficient methods for data processing has always been of interest to researchers in the field of earth sciences. Pattern recognition techniques are appropriate methods for high-dimensional data such as geochemical data. Evaluation of the geochemical distribution of rare earth elements (REEs) requires the use of such methods. In particular, the multivariate nature of REE data makes them a good target for numerical analysis. The main subject of this paper is application of unsupervised pattern recognition approaches in evaluating geochemical distribution of REEs in the Kiruna type magnetite-apatite deposit of Se-Chahun. For this purpose, 42 bulk lithology samples were collected from the Se-Chahun iron ore deposit. In this study, 14 rare earth elements were measured with inductively coupled plasma mass spectrometry (ICP-MS). Pattern recognition makes it possible to evaluate the relations between the samples based on all these 14 features, simultaneously. In addition to providing easy solutions, discovery of the hidden information and relations of data samples is the advantage of these methods. Therefore, four clustering methods (unsupervised pattern recognition)-including a modified basic sequential algorithmic scheme (MBSAS), hierarchical (agglomerative) clustering, k-means clustering and selforganizing map (SOM)-were applied and results were evaluated using the silhouette criterion. Samples were clustered in four types. Finally, the results of this study were validated with geological facts and analysis results from, for example, scanning electron microscopy (SEM), X-ray diffraction (XRD), ICP-MS and optical mineralogy. The results of the k-means clustering and SOM methods have the best matches with reality, with experimental studies of samples and with field surveys. Since only the rare earth elements are used in this division, a good agreement of the results with lithology is considerable. It is concluded that the combination of the proposed methods and geological studies leads to finding some hidden information, and this approach has the best results compared to using only one of them.

Proceedings of the 3rd International Symposium on Movement and Computing - MOCO '16, 2016

In this paper we present a state of the art of the current approaches to visualization of motion capture data. We discuss the data representation, pre-processing techniques, and the design of existing tools and systems. Next we outline the advantages and disadvantages of the systems, some of which are explicitly noted by the original authors. Lastly we conclude with an overall summary and future directions.

npj Climate and Atmospheric Science

East Asia will face a skewed monsoon cycle with soaring flood, drought, and weather whiplash risks in a warming climate. In our objective eight-intraseasonal-monsoon-stage framework, we uncover a ‘dry-get-wetter’ paradigm in East Asia, contesting the fallen ‘rich-get-richer’ common belief. On timing, the Mid-summer and Fall periods are stretching at the expense of three delayed, shortened, and weakened winter stages, especially near the end of the twenty-first century. On threats, entire East Asia will experience up to 14–20 more heavy precipitation days during the rainy Spring to Mid-summer stages. Specifically, the Yangtze basin will suffer from an earlier pluvial period with escalating flood risks. Moreover, societal security and ecosystem resilience in the Huai-Yellow basin, South Japan, and the Korean Peninsula will be challenged by more frequent weather whiplash. Under the monsoon-stage framework, a complete moisture budget decomposition sheds light on the causes of a slower p...

2021

Machine learning is a popular way to find patterns and relationships in high complex datasets. With the nowadays advancements in storage and computational capabilities, some machine-learning techniques are becoming suitable for real-world applications. The aim of this work is to conduct a comparative analysis of machine learning algorithms and conventional statistical techniques. These methods have long been used for clustering large amounts of data and extracting knowledge in a wide variety of science fields. However, the central knowledge of the different methods and their specific requirements for the data set, as well as the limitations of the individual methods, are an obstacle for the correct use of these methods. New machine learning algorithms could be integrated even more strongly into the current evaluation if the right choice of methods were easier to make. In the present work, some different algorithms of machine learning are listed. Four methods (artificial neural netwo...

Musicae Scientiae

In expressive music performance, tempo is known to be a fundamental parameter. In this article, we explored effects of changes in musical tempo on performers’ movement articulations. Eight duos (piano – violin) played two pieces at a predefined tempo, after which this start tempo was gradually increased and decreased. Throughout the different performances, we measured acceleration of the violinists’ head and right wrist, together with the downward force applied by their body to the ground surface. We calculated periodicities in downward force using fast Fourier transform (FFT) analyses and tested whether differences occurred across different tempi. Also, we clustered acceleration and force patterns across different tempi using self-organizing maps (SOMs) and k-means clustering. The results show that a continuous change in performance tempo leads to distinct “performance states” with characteristic bodily behavior in terms of periodic body movement and co-articulated gestures, which ...

Energies, 2022

Recent findings suggest the need for optimal condition monitoring due to increasing counter-productive issues ranging from threats to life, malware, and hardware failures. Several prognostic schemes have been reported across many disciplines; however, the issues of sensor data discrepancy emanating from varying loading and operating conditions of cyber-physical system (CPS) components still remain a challenging factor. Nonetheless, a significant part of these prognostic schemes comprises a sensor/feature fusion module for comprehensive health indicator (HI) construction. This study investigates the prowess of unsupervised fusion algorithms for constructing optimal HI construction on two publicly available datasets—a simulated turbofan engine degradation experiment and an actual production plant condition monitoring dataset. The fusion efficiencies of the algorithms were evaluated using standard metrics for prognostic parameter assessments. The results show that the autoencoder is mo...

Xibei Gongye Daxue Xuebao/Journal of Northwestern Polytechnical University, 2020

Multi-objective optimization can reveal the complex parameter-objective relationships in the high-dimensional design problems. However, the data-extraction and data-presentation of the high-dimensional complex nonlinear system suffers from the increasing dimensionality. Key features and data-distribution of high-dimensional design spaces:parameter and objective spaces could be obtained by using Self-Organizing Maps (SOM) method, which re-clusters the high-dimensional multi-attribute data existing on the Pareto front into several low-dimensional maps. Correlations among all the design variables can be drawn according the colorized topological structure of the maps. Under the constraints including geometric structure and operating parameters, a low-cost and high accurate Kriging surrogate model was established to optimize a hybrid sliding bearing based on the sequential design method. Correlations between 3 objectives:"friction-to-load" ratio, temperature rise, instability t...

Applied Sciences, 2020

Human healthcare is one of the most important topics for society. It tries to find the correct effective and robust disease detection as soon as possible to patients receipt the appropriate cares. Because this detection is often a difficult task, it becomes necessary medicine field searches support from other fields such as statistics and computer science. These disciplines are facing the challenge of exploring new techniques, going beyond the traditional ones. The large number of techniques that are emerging makes it necessary to provide a comprehensive overview that avoids very particular aspects. To this end, we propose a systematic review dealing with the Machine Learning applied to the diagnosis of human diseases. This review focuses on modern techniques related to the development of Machine Learning applied to diagnosis of human diseases in the medical field, in order to discover interesting patterns, making non-trivial predictions and useful in decision-making. In this way, t...

IFIP Advances in Information and Communication Technology, 2019

Managing information security risks in an organisation is one of the most important tasks an organisation has. Unfortunately, due to the complexity of most organisational systems, identifying information security risks can be difficult. One way to identify possible risks in an organisation is to make use of Social Network Analysis (SNA). While they can be used to identify risks, the metrics calculated using SNA are often numerous and daunting to managers unfamiliar with SNA. Furthermore, as the data in this form tend to be uncomfortable to process, educating managers about risks in their organisation can be quite difficult. Also, as these metrics often require quantitative processing in order to be useful, SNA on its own is not always an attractive method to use to identify risks in an organisation. In this paper the use of self-organising maps to identify possible information security risks in an organisation is investigated. Risk data were obtained from an organisation that deals in risk management, which were used to build a social network. A number of metrics associated with risk were calculated from the network, and these metrics were used to cluster the various entities using a self-organising map. Certain entities that pose a possible information security risk were identified. The results suggest that it may be viable to use self-organising maps, in concord with SNA, to more easily identify risks in an organisation using visual methods.

Applied Intelligence, 2018

Multi-objective clustering refers to the partitioning of a given collection of objects into various K-groups based on some similarity/dissimilarity criterion while optimizing different partition quality measures simultaneously. The current paper proposes an automated decomposition based multi-objective clustering technique, SOMDEA clust, which is a fusion of self-organizing map (SOM) and multi-objective differential evolution. A novel reproduction operator is designed where the ensemble of multiple neighborhoods extracted using self-organizing map is used for constructing the variable mating pool size. The probabilities of selecting different sizes of the neighborhood are updated based on their performances in generating new improved solutions in the last few generations. Decomposition based selection scheme is also utilized in our paper which divides the multi-objective optimization (MOO) problem into a number of single objective subproblems. The objective functions corresponding to these subproblems are optimized in a collaborative manner by the use of MOO. The potentiality of the proposed framework is shown for clustering four real-life data sets and five artificial data sets in comparison to some existing multi-objective based clustering techniques, namely MOCK, SMEA clust, MEA clust, a single objective based genetic clustering technique, SOGA and a traditional clustering technique, K-means. To show the utility of SOM based reproduction operators, another decomposition based multi-objective clustering technique (MDEA clust) without the use of SOM based operators is also developed in this paper. In order to show the efficacy of the proposed clustering technique in handling large data sets, two large scale datasets having more than 5000 data points are also utilized. As a real-life application, the proposed clustering technique is applied for scientific/web document clustering where a set of scientific/web documents are partitioned based on their content-similarities. Semantic representation is utilized to covert the text document into a real vector. Experimental results clearly illustrate the effectiveness of fusion of SOM and DE in developing an effective clustering technique.

Cognitive Computation, 2018

Document clustering is the partitioning of a given collection of documents into various Kgroups based on some similarity/dissimilarity criterion. This task has applications in scope detection of journals/conferences, development of some automated peer-review support systems, topic-modeling, latest cognitive-inspired works on text summarization, and classification of documents based on semantics, etc. In the current paper, a cognitive-inspired multi-objective automatic document clustering technique is proposed which is a fusion of self-organizing map (SOM) and multi-objective differential evolution approach. The variable number of cluster centers are encoded in different solutions of the population to determine the number of clusters from a data set in an automated way. These solutions undergo various genetic operations during evolution. The concept of SOM is utilized in designing new genetic operators for the proposed clustering technique. In order to measure the goodness of a clustering solution, two cluster validity indices, Pakhira-Bandyopadhyay-Maulik index, and Silhouette index, are optimized simultaneously. The effectiveness of the proposed approach, namely selforganizing map based multi-objective document clustering technique (SMODoc clust) is shown in automatic classification of some scientific articles and web-documents. Different representation schemas including tf, tf-idf and word-embedding are employed to convert articles in vector-forms. Comparative results with respect to internal cluster validity indices, namely, Dunn index and Davies-Bouldin index, are shown against several state-of-the-art clustering techniques including three multiobjective clustering techniques namely MOCK, VAMOSA, NSGA-II-Clust, single objective genetic algorithm (SOGA) based clustering technique, K-means, and single-linkage clustering. Results obtained clearly show that our approach is better than existing approaches. The validation of the obtained results is also shown using statistical significant t tests.

Applied Intelligence, 2019

In the current paper, a novel approach is proposed for bi-clustering of gene expression data using the fusion of differential evolution framework and self-organizing map (SOM), named as BiClustSMEA. Variable number of gene and condition cluster centers are encoded in different solutions of the population to determine the number of bi-clusters from a dataset in an automated way. The concept of SOM is utilized in designing new genetic operators for both gene and condition clusters to reach to the optimal solution in a faster way. In order to measure the goodness of a bi-clustering solution, three bi-cluster quality measures, mean squared error, row variance, and bi-cluster size, are optimized simultaneously using differential evolution as the underlying optimization strategy. The concept of polynomial mutation is incorporated in our framework to generate highly diverse solutions which in turn helps in faster convergence. The proposed approach is applied on two reallife microarray gene expression datasets and results are compared with various state-of-the-art techniques. Results obtained clearly illustrate that our approach extracts high-quality bi-clusters as compared to other methods and also it converges much faster than other competitors. Further, the obtained results are validated using statistical significance test and biological significance test.

Cognitive Computation, 2010

This paper presents a tree evolutionary recurrent self-organizing model, based on locally adapting neighborhood radii and multiple prototype vectors named GH-Ad-RSOM. It is a variant of the Growing Hierarchical Self-Organizing Map GHSOM. The proposed GHSOM variant is characterized by a hierarchical model, composed of independent RSOMs (many RSOM), based on a locally adapting neighborhood radii and multiple prototype vectors. The method shows better robustness of GH-Ad-RSOM and high vowel classification rates compared to classic GHSOM and other variant of GHSOM.

Adaptive Behavior, 2018

Embodied Cognitive Robotics focuses its attention on the design of artificial agents capable of performing cognitive tasks autonomously. A central issue in this consists in studying process by which agents learn through interaction with their environment. Embodied Cognitive Robotics aims to implement models of cognitive processes coming from Cognitive Sciences. The guidelines in this research area are a direct response to the shortcomings of Classical Artificial Intelligence, where high-level tasks and behaviors were studied. This article describes the work carried out in the Cognitive Robotics Laboratory at the Universidad Autónoma del Estado de Morelos (UAEM). Our work is based on the concept of low-level sensorimotor schemes coded by Internal Models, thus falling as a matter of course within the tenets of Embodied Cognition, particularly with the idea that cognition must be understood as occurring in agents that have a body with which they interact in a specific environment. It i...

Frontiers in Computer Science, 2020

Autonomous Robots, 2018

The paper presents a neurorobotics cognitive model explaining the understanding and generalisation of nouns and verbs combinations when a vocal command consisting of a verb-noun sentence is provided to a humanoid robot. The dataset used for training was obtained from object manipulation tasks with a humanoid robot platform; it includes 9 motor actions and 9 objects placing placed in 6 different locations), which enables the robot to learn to handle real-world objects and actions. Based on the multiple timescale recurrent neural networks, this study demonstrates its generalisation capability using a large data-set, with which the robot was able to generalise semantic representation of novel combinations of noun-verb sentences, and therefore produce the corresponding motor behaviours. This generalisation process is done via the grounding process: different objects are being interacted, and associated, with different motor behaviours, following a learning approach inspired by developmental language acquisition in infants. Further analyses of the learned network dynamics and representations also demonstrate how the generalisation is possible via the exploitation of this functional hierarchical recurrent network. Keywords Recurrent artificial neural networks • Language learning • Multiple timescale recurrent neural network • Developmental robotics • Neurorobotics B Junpei Zhong

Journal of Imaging, 2020

With the increase in life expectancy, one of the most important topic for scientific research, especially for the elderly, is good nutrition. In particular, with an advanced age and health issues because disorders such as Alzheimer and dementia, monitoring the subjects’ dietary habits to avoid excessive or poor nutrition is a critical role. Starting from an application aiming to monitor the food intake actions of people during a meal, already shown in a previously published paper, the present work describes some improvements that are able to make the application work in real time. The considered solution exploits the Kinect v1 device that can be installed on the ceiling, in a top-down view in an effort to preserve privacy of the subjects. The food intake actions are estimated from the analysis of depth frames. The innovations introduced in this document are related to the automatic identification of the initial and final frame for the detection of food intake actions, and to the str...

Remote Sensing, 2014

Ecosystem state can be characterized by a set of attributes that are related to the ecosystem functionality, which is a relevant issue in understanding the quality and quantity of ecosystem services and goods, adaptive capacity and resilience to perturbations. This study proposes a major identification of Ecosystem Functional Types (EFTs) in Spain to characterize the patterns of ecosystem functional diversity and status, from several functional attributes as the Normalized Difference Vegetation Index (NDVI), Land Surface Temperature (LST) and Albedo. For this purpose, several metrics, related to the spatial variability in seasonal and annual patterns (e.g., relative range), have been derived from remote sensing time series of 1 km MODIS over the period 2000-2009. Moreover, precipitation maps from data provided by the AEMet (Agencia Estatal de Meteorología) and the corresponding aridity and humidity indices were also included in the analysis. To create the EFTs, the potential of the joint use of Kohonen's Self-Organizing Map (SOM) and the k-means clustering algorithm was tested. The EFTs were analyzed using different remote sensing (i.e., Gross Primary Production) and climatic variables. The relationship of the EFTs with existing land cover datasets and climatic data were analyzed through a correspondence analysis (CA). The trained SOM have shown feasible in providing a comprehensive view on the functional attributes patterns and a remarkable potential for the quantification of

Granular Computing, 2016

We present our results on the adoption of a settheoretic framework for granular computing to situation awareness. The proposed framework guarantees a high degree of flexibility in the process of creation of granules and granular structures allowing to satisfy the wide variety of requirements for perception and comprehension of situations where some elements must be perceived per similarity, others per spatial proximity, some must be fused to improve their comprehension, and so on. A second value is the support for approximate reasoning in situation awareness. A granular structure in particular represents a snapshot of a situation, and is a building block for the development of tools and techniques to reason on situation in order to reduce situation awareness errors and accelerate the process of decision-making. To this purpose, we show a technique to support operators in the analysis of conformity between a recognized situation and an expected one. A third value is the fact that we can support operators in having rapid and indicative measures of how two situations, e.g. a recognized and a projected, may differ. A preliminary evaluation instantiating our approach with selforganizing maps is reported and discussed. The results are encouraging with respect to the capability of improving perception and comprehension of a situation, reducing comprehension errors and supporting projection of situations.

Transport and Telecommunication Journal, 2020

This paper deals with an analysis of Kohonen map usage possibility for real-time evaluation of end-user video quality perception. The Quality of Service framework (QoS) describes how the network impairments (network utilization or packet loss) influence the picture quality, but it does not reflect precisely on customer subjective perceived quality of received video stream. There are several objective video assessment metrics based on mathematical models trying to simulate human visual system but each of them has its own evaluation scale. This causes a serious problem for service providers to identify a critical point when intervention into the network behaviour is needed. On the other hand, subjective tests (Quality of Experience concept) are time-consuming and costly and of course, cannot be performed in real-time. Therefore, we proposed a mapping function able to predict subjective end-user quality perception based on the situation in a network, video stream features and results o...

Environmental Health Insights, 2017

The abundance of airborne particulate matter with an aerodynamic equivalent diameter of 2.5 µm or less (PM2.5) is a significant environmental and health issue. Many tools have been used to examine the relationship between PM2.5 abundance and meteorological variables, but some of the relationships are nonlinear, non-Gaussian, and even unknown. Machine learning provides a broad range of practical algorithms to help examine this issue. In this study, we use machine learning to classify the morphology of PM2.5 seasonal cycles in East Asia. Machine learning is able to objectively classify the seasonal cycles and, without a priori assumption, is able to clearly distinguish between urban and rural areas. We show an example of this in the Sichuan Basin of China. Furthermore, machine learning is also able to provide physical insights by identifying the key factors associated with each distinct shape of the seasonal cycle, such as highlighting the key role played by the topography and the bui...

Journal of the Nigerian Society of Physical Sciences

Self-organizing maps (SOMs) and other artificial intelligence approaches developed by Kohonen can be used to model and solve environmental challenges. To emphasize the classification of Physico-chemical parameters of the Inaouen watershed, we presented a classification strategy based on a self-organizing topological map (SOM) artificial neural network in this study. The use of a self-organizing map to classify samples resulted in the following five categories: Low quantities of Sodium Na (mg/l), Potassium k(mg/l), Magnesium Mg(mg/l), Calcium Ca(mg/l), Sulfates SO4(mg/l), and Total Dissolved Solids TDS (mg/l) distinguish Classes 2 and 3. Bicarbonate HCO3 (mg/l), Total Dissolved Solids TDS (mg/l), Total Alkalinity CaCO3(mg/l), Mg(mg/l), Calcium Ca (mg/l), and electrical conductivity Cond (ms/cm) are slightly greater in Classes 1 and 4. Except for Dissolved Oxygen D.O. (mg/l) and Nitrate NO3(mg/l), Class 5 has exceptionally high values for all metrics. The results suggest that Kohonen&...

Global Biogeochemical Cycles, 2017

In this study we used fluorescence excitation and emission matrix spectroscopy, hydrographic data, and a self-organizing map (SOM) analysis to assess the spatial distribution of labile and refractory fluorescent dissolved organic matter (FDOM) for the Chukchi and Beaufort Seas at the time of a massive under-ice phytoplankton bloom during early summer 2011. Biogeochemical properties were assessed through decomposition of water property classes and sample classification that employed a SOM neural network-based analysis which classified 10 clusters from 269 samples and 17 variables. The terrestrial, humic-like component FDOM (ArC1, 4.98 ± 1.54 Quinine Sulfate Units (QSU)) and protein-like component FDOM (ArC3, 1.63 ± 0.88 QSU) were found to have elevated fluorescence in the Lower Polar Mixed Layer (LPML) (salinity~29.56 ± 0.76). In the LPML water mass, the observed contribution of meteoric water fraction was 17%, relative to a 12% contribution from the sea ice melt fraction. The labile ArC3-protein-like component (2.01 ± 1.92 QSU) was also observed to be elevated in the Pacific Winter Waters mass, where the under-ice algal bloom was observed (~40-50 m). We interpreted these relationships to indicate that the accumulation and variable distribution of the protein-like component on the shelf could be influenced directly by sea ice melt, transport, and mixing processes and indirectly by the in situ algal bloom and microbial activity. ArC5, corresponding to what is commonly considered marine humic FDOM, indicated a bimodal distribution with high values in both the freshest and saltiest waters. The association of ArC5 with deep, dense salty water is consistent with this component as refractory humic-like FDOM, whereas our evidence of a terrestrial origin challenges this classic paradigm for this component. During the "Impacts of Climate Ecosystems and Chemistry of the Arctic Pacific Environment" (ICESCAPE) 2011 research cruise, an under-ice bloom [Arrigo et al., 2012] was observed concurrent with the increasing coverage of melt-ponds present on annual sea ice, which were found to increase light transmission in the subsurface waters fourfold compared to adjacent snow-free ice [Frey et al., 2011; Arrigo et al., 2012; Arrigo and van Dijken, 2015]. The recent dramatic decline in multiyear sea ice is thought to be a result from the heat flux from the Pacific inflow through the Bering Strait [Woodgate et al., 2010]. Knowledge of how the under-ice algal bloom and sea ice melt affect the distribution of organic matter constituents with MENDOZA ET AL.

Brazilian Journal of Pharmaceutical Sciences, 2011

Tissue damage due to oxidative stress is directly linked to development of many, if not all, human morbidity factors and chronic diseases. In this context, the search for dietary natural occurring molecules with antioxidant activity, such as flavonoids, has become essential. In this study, we investigated a set of 41 flavonoids (23 flavones and 18 flavonols) analyzing their structures and biological antioxidant activity. The experimental data were submitted to a QSAR (quantitative structure-activity relationships) study. NMR 13C data were used to perform a Kohonen self-organizing map study, analyzing the weight that each carbon has in the activity. Additionally, we performed MLR (multilinear regression) using GA (genetic algorithms) and molecular descriptors to analyze the role that specific carbons and substitutions play in the activity.

Sustainability, 2022

The purpose of this research was to derive promising technologies for the transport of hydrogen fuel cells, thereby supporting the development of research and development policy and presenting directions for investment. We also provide researchers with information about technology that will lead the technology field in the future. Hydrogen energy, as the core of carbon neutral and green energy, is a major issue in changing the future industrial structure and national competitive advantage. In this study, we derived promising technology at the core of future hydrogen fuel cell transportation using the published US patent and paper databases (DB). We first performed text mining and data preprocessing and then discovered promising technologies through generative topographic mapping analysis. We analyzed both the patent DB and treatise DB in parallel and compared the results. As a result, two promising technologies were derived from the patent DB analysis, and five were derived from the...

Studies in Computational Intelligence, 2020

Although common vulnerabilities and exposures data (CVE) is commonly known and used to keep vulnerability descriptions. It lacks enough classifiers that increase its usability. This results in focusing on some well-known vulnerabilities and leaving others during the security tests. Better classification of this dataset would result in finding solutions to a larger set of vulnerabilities/exposures. In this research, vulnerability and exposure data (CVE) is examined in detail using both manual and computerized content analysis techniques. Later, graph theoretical techniques are used to scrutinize the CVE data. The computerized content analysis made it possible to find out 94 concepts associated with the CVE records. The author was able to relate these concepts to 11 logical groups. Using the network of the relationships of these 94 concepts further in the graph theoretical analysis made it possible to discover groups of contents, thus, the CVE items which have similarities. Moreover, lacking some concepts pointed out the problems related to CVE such as delays in the review CVE process or not being preferred by some user groups.

Frontiers in Computational Neuroscience, 2021

Recurrent neural networks with associative memory properties are typically based on fixed-point dynamics, which is fundamentally distinct from the oscillatory dynamics of the brain. There have been proposals for oscillatory associative memories, but here too, in the majority of cases, only binary patterns are stored as oscillatory states in the network. Oscillatory neural network models typically operate at a single/common frequency. At multiple frequencies, even a pair of oscillators with real coupling exhibits rich dynamics of Arnold tongues, not easily harnessed to achieve reliable memory storage and retrieval. Since real brain dynamics comprises of a wide range of spectral components, there is a need for oscillatory neural network models that operate at multiple frequencies. We propose an oscillatory neural network that can model multiple time series simultaneously by performing a Fourier-like decomposition of the signals. We show that these enhanced properties of a network of H...

Geochemistry: Exploration, Environment, Analysis, 2007

Introduction: Trace elements in groundwaters can be concentrated to unacceptable levels during evaporation of saline groundwaters into waste destined for disposal. In addition, saline waters provide a host for sulfate-reducing bacteria which can also lead to build-up of toxic metals in pond sediments. Because evaporation ponds are either in place or proposed as part of the solution to Australia's groundwater salinity problems, we have been evaluating the potential for trace metal-related problems. For example, in the San Joaquin Valley, in California USA, such problems have resulted in the cessation of use of evaporation ponds due to high concentrations of Se, Mo and U in the sediments of the ponds and subsequent detrimental effects on wild fowl. In Australia, concentrations of U and Ra (radium) around salt-lakes have misled companies exploring for U into unwarranted expenditure. Groundwater data sets with the complete major cation and anion chemistry are widely available but data on toxic trace elements such as As, Cr, F, Se, U are much rarer. The study reported here investigated the use of data imputation methods to extend the small database of samples with multi-element analyses using the larger database of incomplete databases.

Handbook of Materials Modeling, 2020

Accurate prediction of airflow distribution and aerosol transport in the human lungs, which are difficult to be measured in vivo but important to understand the structure and function relationship, is challenging. It is because the interplay between them spans more than two orders of magnitude in dimension from the trachea to alveoli. This chapter reviews the techniques and strategies for modeling lungs both within and between subjects, viz., subject specificity versus generalization from individuals to populations, with both exhibiting multiscale characteristics. For "within-subject" modeling, a computed tomography (CT)derived subject-specific computational fluid dynamics (CFD) lung model is presented. The pipeline for building such an imaging-based lung model is composed of image segmentation and processing, geometrical modeling labeled with anatomical information, image registration, three-dimensional (accurate) and one-dimensional (approximate) coupling techniques, and a high-fidelity turbulent flow model. The subject-specific model is essential in predicting local structural and functional interactions. For "between subjects" modeling, machine learning is employed to identify homogeneous subpopulations (clusters), among healthy and diseased populations, aiming to bridge individual and population scales. For this purpose, three major issues need to be addressed. They are intersubject variability (due to, e.g., gender, age, and height), inter-site variability (due to scanner and imaging protocol differences), and definition of quantitative CT imaging-based metrics at multiple scales (due to alterations at different disease stages) needed for clustering. The use of the cluster membership to select representative subjects for detailed CFD analysis enables an examination of the cluster-specific structural and functional relationships.

Minerals

Natural or artificial light allows us to see and analyze matter with our eyes, which are the first tools used in several experiments. In geosciences, particularly in mineralogy, light is used for optical microscopy observations. Reflected and transmitted light applied to the study of ore deposits can be useful to discriminate between gangue from precious phases. Knowledge of the structural and morphological characteristics, combined with the quantitative evaluation of mineral abundance, is fundamental for determining the grade of ore deposits. The accuracy and reliability of the information are closely linked to the ability of the mineralogist, who more and more often uses Scanning Electron technology and automated mineralogy systems to validate the observations or solve complex mineralogy. While highly accurate, these methods are often prohibitively expensive. The use of image analysis using standard algorithms and artificial intelligence, available as open source, and commercial p...

Water

Shallow tidal creek systems or triblets are often overlooked when documenting and measuring the spatial extent of pollutants of emerging concern despite much of the population living in and around these areas. An innovative in situ fluorometric instrument coupled with a Self-Organi21zing Map was utilized in Chesapeake Bay’s mesohaline tidal creek system to analyze CDOM, dissolved oxygen, optical brighteners, and oils. The in situ fluorometer proved helpful as a rapid reconnaissance tool complementing the investigation when attached to a CTD instrument. This baseline research showed that CDOM follows non-conservative properties in spring and more conservative behavior in the fall. The results show that the Self-Organizing Map method is a suitable alternative to traditional statistical techniques and may be better at finding key patterns that might otherwise have been obscured by high variability. For example, oils revealed a pattern with residual runoff from highways or boating, whil...

Heritage

In this paper, we discuss a new approach to the analysis of multi/hyper-spectral data sets, based on the Interesting Features Finder (IFF) method. The IFF is a simple algorithm recently proposed in the framework of Laser-Induced Breakdown Spectroscopy (LIBS) spectral analysis for detecting ‘interesting’ spectral features independently of the variance they represent in a set of spectra. To test the usefulness of this method to multispectral analysis, we show in this paper the results of its application on the recovery of a ‘lost’ painting from the Etruscan hypogeal tomb of the Volumni (3rd century BCE—1st century CE) in Perugia, Italy. The results obtained applying the IFF algorithm are compared with the results obtained by applying Blind Source Separation (BSS) techniques and Self-Organized Maps (SOM) to a multispectral set of 17 fluorescence and reflection images. From this comparison emerges the possibility of using the IFF algorithm to obtain rapidly and simultaneously, by varyin...

Metabolites, 2020

As researchers are increasingly able to collect data on a large scale from multiple clinical and omics modalities, multi-omics integration is becoming a critical component of metabolomics research. This introduces a need for increased understanding by the metabolomics researcher of computational and statistical analysis methods relevant to multi-omics studies. In this review, we discuss common types of analyses performed in multi-omics studies and the computational and statistical methods that can be used for each type of analysis. We pinpoint the caveats and considerations for analysis methods, including required parameters, sample size and data distribution requirements, sources of a priori knowledge, and techniques for the evaluation of model accuracy. Finally, for the types of analyses discussed, we provide examples of the applications of corresponding methods to clinical and basic research. We intend that our review may be used as a guide for metabolomics researchers to choose ...

Annals of Clinical and Translational Neurology, 2019

Background: The selection of optimal deep brain stimulation (DBS) parameters is time-consuming, experience-dependent, and best suited when acute effects of stimulation can be observed (e.g., tremor reduction). Objectives: To test the hypothesis that optimal stimulation location can be estimated based on the cortical connections of DBS contacts. Methods: We analyzed a cohort of 38 patients with Parkinson's disease (24 training, and 14 test cohort). Using whole-brain probabilistic tractography, we first mapped the cortical regions associated with stimulation-induced efficacy (rigidity, bradykinesia, and tremor improvement) and side effects (paresthesia, motor contractions, and visual disturbances). We then trained a support vector machine classifier to categorize DBS contacts into efficacious, defined by a therapeutic window ≥2 V (threshold for side effect minus threshold for efficacy), based on their connections with cortical regions associated with efficacy versus side effects. The connectivitybased classifications were then compared with actual stimulation contacts using receiver-operating characteristics (ROC) curves. Results: Unique cortical clusters were associated with stimulation-induced efficacy and side effects. In the training dataset, 42 of the 47 stimulation contacts were accurately classified as efficacious, with a therapeutic window of ≥3 V in 31 (66%) and between 2 and 2.9 V in 11 (24%) electrodes. This connectivity-based estimation was successfully replicated in the test cohort with similar accuracy (area under ROC = 0.83). Conclusions: Cortical connections can predict the efficacy of DBS contacts and potentially facilitate DBS programming. The clinical utility of this paradigm in optimizing DBS outcomes should be prospectively tested, especially for directional electrodes.

Logic Journal of the IGPL, 2018

This research proposes the analysis and subsequent characterization of Android malware families, by means of low dimensional visualizations using dimensional reduction techniques. The well-known Malgenome dataset, coming from the Android Malware Genome Project, has been thoroughly analysed through six dimensionality reduction techniques: Principal Component Analysis, Maximum Likelihood Hebbian Learning, Cooperative Maximum Likelihood Hebbian Learning, Curvilinear Component Analysis, Isomap and Self Organizing Map. Results obtained enable a clear visual analysis of the structure of this high-dimensionality dataset, letting us gain deep knowledge about the nature of such Android malware families. Interesting conclusions are obtained from the real-life dataset under analysis.

Journal of Nondestructive Evaluation, Diagnostics and Prognostics of Engineering Systems, 2021

A novel failure model updating methodology is presented in this paper for composite materials. The innovation in the approach presented is found in both the experimental and computational methods used. Specifically, a dominant bottleneck in data-driven failure model development relates to the types of data inputs that could be used for model calibration or updating. To address this issue, nondestructive evaluation data obtained while performing mechanical testing at the laboratory scale are used in this paper to form a damage metric based on a series of processing steps that leverage raw sensing inputs and provide progressive failure curves that are then used to calibrate the damage initiation point computed by full-field three-dimensional finite element simulations of fiber-reinforced composite material that take into account both intra- and interlayer damage. Such curves defined based on nondestructive evaluation data are found to effectively monitor the progressive failure proces...

SPIE Proceedings, 2017

High-rate systems operating in the 10 μs to 10 ms timescale are likely to experience damaging effects due to rapid environmental changes (e.g., turbulence, ballistic impact). Some of these systems could benefit from real-time state estimation to enable their full potential. Examples of such systems include blast mitigation strategies, automotive airbag technologies, and hypersonic vehicles. Particular challenges in high-rate state estimation include: 1) complex time varying nonlinearities of system (e.g. noise, uncertainty, and disturbance); 2) rapid environmental changes; 3) requirement of high convergence rate. Here, we propose using a Variable Input Observer (VIO) concept to vary the input space as the event unfolds. When systems experience high-rate dynamics, rapid changes in the system occur. To investigate the VIO's potential, a VIObased neuro-observer is constructed and studied using experimental data collected from a laboratory impact test. Results demonstrate that the input space is unique to different impact conditions, and that adjusting the input space throughout the dynamic event produces better estimations than using a traditional fixed input space strategy.

Structural Control and Health Monitoring, 2018

High-rate dynamic systems are defined as systems being exposed to highly dynamic environments that comprise high-rate and high-amplitude events. Examples of such systems include civil structures exposed to blast, space shuttles prone to debris strikes, and aerial vehicles experiencing in-flight changes. The high-rate dynamic characteristics of these systems provides several possibilities for state estimators to improve performance, including a high potential to reduce injuries and save lives. In this paper, opportunities and challenges that are specific to state estimation of high-rate dynamic systems are presented and discussed. It is argued that a possible path to design of state estimators for high-rate dynamics is the utilization of adaptive data-based observers but that further research needs to be conducted to increase their convergence rate. An adaptive neuro-observer is designed to examine the particular challenges in selecting an appropriate input space in high-rate state estimation. It is found that the choice of inputs has a significant influence on the observer performance for high-rate dynamics when compared against a low-rate environment. Additionally, misrepresentation of a system dynamics through incorrect input spaces produces large errors in the estimation, which could potentially trick the decision-making process in a closed-loop system in making bad judgments.

Neural Computing and Applications, 2018

Engineering systems experiencing events of amplitudes higher than 100 g n for a duration under 100 ms, here termed high-rate dynamics, can undergo rapid damaging effects. If the structural health of such systems could be accurately estimated in a timely manner, preventative measures could be employed to minimize adverse effects. For complex high-rate problems, adaptive observers have shown promise due to their capability to deal with nonstationary, noisy, and uncertain systems. However, adaptive observers have slow convergence rates, which impede their applicability to the high-rate problems. To improve on the convergence rate, we propose a variable input space concept for optimizing the use of data history of high-rate dynamics, with the objective to produce an optimal representation of the system of interest. Using the embedding theory, the algorithm sequentially selects and adapts a vector of inputs that preserves the essential dynamics of the high-rate system. In this paper, the variable input space is integrated in a wavelet neural network, which constitutes a variable input observer. The observer is simulated using experimental data from a high-rate system. Different input space adaptation methods are studied, and the performance is also compared against an optimized fixed input strategy. It is found that a smooth transition of the input space eliminates error spikes and yields faster convergence. The variable input observer is further studied in a hybrid model-/data-driven formulation, and results demonstrate significant improvement in performance gained from the added physical knowledge.

Frontiers of Computer Science, 2015

Human action recognition from skeletal data is an important and active area of research in which the state of the art has not yet achieved near-perfect accuracy on many wellknown datasets. In this paper, we introduce the Distribution of Action Movements Descriptor, a novel action descriptor based on the distribution of the directions of the motions of the joints between frames, over the set of all possible motions in the dataset. The descriptor is computed as a normalized histogram over a set of representative directions of the joints, which are in turn obtained via clustering. While the descriptor is global in the sense that it represents the overall distribution of movement directions of an action, it is able to partially retain its temporal structure by applying a windowing scheme. The descriptor, together with a standard classifier, outperforms several state-of-the-art techniques on many wellknown datasets.

Nuclear Fusion, 2014

Neural Computing and Applications, 2020

In communication networks resilience or structural coherency, namely the ability to maintain total connectivity even after some data links are lost for an indefinite time, is a major design consideration. Evaluating resilience is a computationally challenging task since it often requires examining a prohibitively high number of connections or of node combinations, depending on the structural coherency definition. In order to study resilience, communication systems are treated in an abstract level as graphs where the existence of an edge depends heavily on the local connectivity properties between the two nodes. Once the graph is derived, its resilience is evaluated by a tensor stack network (TSN). TSN is an emerging deep learning classification methodology for big data which can be expressed either as stacked vectors or as matrices, such as images or oversampled data from multiple-input and multiple-output digital communication systems. As their collective name suggests, the architecture of TSNs is based on tensors, namely higher-dimensional vectors, which simulate the simultaneous training of a cluster of ordinary multilayer feedforward neural networks (FFNNs). In the TSN structure the FFNNs are also interconnected and, thus, at certain steps of the training process they learn from the errors of each other. An additional advantage of the TSN training process is that it is regularized, resulting in parsimonious classifiers. The TSNs are trained to evaluate how resilient a graph is, where the real structural strength is assessed through three established resiliency metrics, namely the Estrada index, the odd Estrada index, and the clustering coefficient. Although the approach of modelling the communication system exclusively in structural terms is function oblivious, it can be applied to virtually any type of communication network independently of the underlying technology. The classification achieved by four configurations of TSNs is evaluated through six metrics, including the F1 metric as well as the type I and type II errors, derived from the corresponding contingency tables. Moreover, the effects of sparsifying the synaptic weights resulting from the training process are explored for various thresholds. Results indicate that the proposed method achieves a very high accuracy, while it is considerably faster than the computation of each of the three resilience metrics. Concerning sparsification, after a threshold the accuracy drops, meaning that the TSNs cannot be further sparsified. Thus, their training is very efficient in that respect. Keywords Tensor stack network Á Tensor algebra Á Deep learning Á Big data Á Higher-order data Á Graph mining Á Graph resilience Á Estrada index Á Clustering coefficient Á Multilinear classification Á Sparsification Á Regularization Á TensorFlow Á Keras Mathematics Subject Classification 05C50 Á 05C62 Á 05C76 Á 68R10 Á 94C15 Á 97K30 Neural Computing and Applications

IGI Global eBooks, 2018

Mathematical Geosciences, 2018

In many fields such as robotics (Poppinga et al. 2008), computer vision (Mitra and Nguyen 2003), digital photogrammetry (Yang and Förtsner 2010), surface reconstruction (Nurunnabi et al. 2012), computational geometry (Lukács et al. 1998) as well as in the increasing applications of laser scanning (Stathas et al. 2003), it is a fundamental task for extracting features from 3D point cloud. Since the physical limitations of the sensors, the occlusions, multiple reflectance and noise can produce off-surface points, robust fitting techniques are required. Robust fitting means an estimation technique which is able to estimate accurate model parameters not only despite small-scale noise in the data set but occasionally large scale measurement errors (outliers). Outliers definition is not easy. Perhaps considering the problem from the practical point of the view, we can say that data points, which appearance in the data set causes dramatically change in the result of the parameter estimation can be labeled as outliers. Basically there are two different methods to handle outliers: (a) weighting out outliers (b) discarding outliers Weighting outliers means that we do not kick out certain data points labeled as outlier but during the parameter estimation process we take them with a low weight into consideration in the objective function. Such a technique is the good old Danish method. The other technique will try to identify data points, which make "troubles" during the parameter estimation process. Troubles mean that their existence in the data set

Ore Geology Reviews, 2020

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Journal of Visualization, 2016

In a variety of research and application areas, graphs are an important structure for data modeling and analysis. While graph properties can have a crucial influence on the performance of graph algorithms, and thus on the outcome of experiments, often only basic analysis of the graphs under investigation in an experimental evaluation is performed and a few characteristics are reported in publications. We present Graph Landscape, a concept for the visual analysis of graph set properties. The Graph Landscape aims to support researchers to explore graphs and graph sets regarding their properties, to allow to select good experimental test sets, analyze newly generated sets, compare sets and assess the validity (or range) of experimental results and corresponding conclusions. Keywords Graphs Á Multidimensional data Á Visualization Á Analysis 1 Introduction Graphs are a versatile data structure that is used to model data in a variety of domains, including biology, security, social sciences, telecommunication, economics, and engineering. The resulting graphs and their properties facilitate the analysis of the data, e.g., in biology to find out the underlying principles of biological processes, or in social sciences to discover leaders and dynamic group behavior. Many online databases and collections of graphs are available that support such network analysis (KEGG Kyoto

Environmental Monitoring and Assessment, 2019

PM 2.5 air pollution is a significant issue for human health all over the world, especially in East Asia. A large number of ground-based measurement sites have been established over the last decade to monitor real-time PM 2.5 concentration. However, even this enhanced observational network leaves many gaps in characterizing the PM 2.5 spatial distribution. Machine learning provides a variety of algorithms to help deal with these large spatial gaps-combining both remotely sensed and in situ observation data to estimate the global PM 2.5 concentration. This study used a PM 2.5 data product of six regions from the results of an unsupervised self-organizing map (SOM) with optimized ensemble learning approaches to highlight the most important meteorological and surface variables associated with PM 2.5 concentration. These variables were then examined via multiple linear regression models to provide physical mechanistic insights into the morphology of the PM 2.5 annual cycles.

Environmental Monitoring and Assessment, 2015

Carbon capture and storage (CCS) is gaining interest as a significant global option to reduce emissions of CO 2. CCS development requires an assessment of the potential risks associated with CO 2 leakages from storage sites. Laboratory leaching tests have proved to be a useful tool to study the potential mobilization of metals from contaminated sediment in a decreased-pH environment that mimics such a leakage event. This work employs a Self-Organizing Map (SOM) tool to interpret and analyze the release of Dissolved Organic Carbon (DOC), As, Cd, Cr, Cu, Ni, Pb and Zn from equilibrium, column and pH-dependent leaching tests. In these tests, acidified seawater is used for simulating different CO 2 leakage scenarios. Classification was carried out detailing the mobilization of contaminants for environments of varying pH, liquid-to-solid ratio and type of contact of the laboratory leaching tests. Component planes in the SOMs allow visualization of the results and the determination of the worst-case of element-release. The pH-dependent leaching test with initial addition of either base or acid was found to mobilize the highest concentrations of metals.

Frontiers in Computational Neuroscience, 2018

Oscillatory phenomena are ubiquitous in the brain. Although there are oscillator-based models of brain dynamics, their universal computational properties have not been explored much unlike in the case of rate-coded and spiking neuron network models. Use of oscillator-based models is often limited to special phenomena like locomotor rhythms and oscillatory attractor-based memories. If neuronal ensembles are taken to be the basic functional units of brain dynamics, it is desirable to develop oscillator-based models that can explain a wide variety of neural phenomena. Autoencoders are a special type of feed forward networks that have been used for construction of large-scale deep networks. Although autoencoders based on rate-coded and spiking neuron networks have been proposed, there are no autoencoders based on oscillators. We propose here an oscillatory neural network model that performs the function of an autoencoder. The model is a hybrid of rate-coded neurons and neural oscillators. Input signals modulate the frequency of the neural encoder oscillators. These signals are then multiplexed using a network of rate-code neurons that has afferent Hebbian and lateral anti-Hebbian connectivity, termed as Lateral Anti Hebbian Network (LAHN). Finally the LAHN output is de-multiplexed using an output neural layer which is a combination of adaptive Hopf and Kuramoto oscillators for the signal reconstruction. The Kuramoto-Hopf combination performing demodulation is a novel way of describing a neural phase-locked loop. The proposed model is tested using both synthetic signals and real world EEG signals. The proposed model arises out of the general motivation to construct biologically inspired, oscillatory versions of some of the standard neural network models, and presents itself as an autoencoder network based on oscillatory neurons applicable to time series signals. As a demonstration, the model is applied to compression of EEG signals.

Science Advances, 2019

Multidecadal "megadroughts" were a notable feature of the climate of the American Southwest over the Common era, yet we still lack a comprehensive theory for what caused these megadroughts and why they curiously only occurred before about 1600 CE. Here, we use the Paleo Hydrodynamics Data Assimilation product, in conjunction with radiative forcing estimates, to demonstrate that megadroughts in the American Southwest were driven by unusually frequent and cold central tropical Pacific sea surface temperature (SST) excursions in conjunction with anomalously warm Atlantic SSTs and a locally positive radiative forcing. This assessment of past megadroughts provides the first comprehensive theory for the causes of megadroughts and their clustering particularly during the Medieval era. This work also provides the first paleoclimatic support for the prediction that the risk of American Southwest megadroughts will markedly increase with global warming.

Lecture Notes in Computer Science, 2007

This multidisciplinary study focuses on the application and comparison of several topology preserving mapping models upgraded with some classifier ensemble and boosting techniques in order to improve those visualization capabilities. The aim is to test their suitability for classification purposes in the field of food industry and more in particular in the case of dry cured ham. The data is obtained from an electronic device able to emulate a sensory olfative taste of ham samples. Then the data is classified using the previously mentioned techniques in order to detect which batches have an anomalous smelt (acidity, rancidity and different type of taints) in an automated way.

Chemical Engineering Journal, 2021

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

2022 IEEE Symposium Series on Computational Intelligence (SSCI)

Modern Artificial Intelligence (AI) enabled Intrusion Detection Systems (IDS) are complex black boxes. This means that a security analyst will have little to no explanation or clarification on why an IDS model made a particular prediction. A potential solution to this problem is to research and develop Explainable Intrusion Detection Systems (X-IDS) based on current capabilities in Explainable Artificial Intelligence (XAI). In this paper, we create a Self Organizing Maps (SOMs) based X-IDS system that is capable of producing explanatory visualizations. We leverage SOM's explainability to create both global and local explanations. An analyst can use global explanations to get a general idea of how a particular IDS model computes predictions. Local explanations are generated for individual datapoints to explain why a certain prediction value was computed. Furthermore, our SOM based X-IDS was evaluated on both explanation generation and traditional accuracy tests using the NSL-KDD and the CIC-IDS-2017 datasets.