Multidimensional Network Research Articles

Urdu Nastaliq recognition using convolutional–recursive deep learning

Recent developments in recognition of cursive scripts rely on implicit feature extraction methods that provide better results as compared to traditional hand-crafted feature extraction approaches. We present a hybrid approach based on explicit feature extraction by combining convolutional and recursive neural networks for feature learning and classification of cursive Urdu Nastaliq script. The first layer extracts low-level translational invariant features using Convolutional Neural Networks (CNN) which are then forwarded to Multi-dimensional Long Short-Term Memory Neural Networks (MDLSTM) for contextual feature extraction and learning. Experiments are carried out on the publicly available Urdu Printed Text-line Image (UPTI) dataset using the proposed hierarchical combination of CNN and MDLSTM. A recognition rate of up to 98.12% for 44-classes is achieved outperforming the state-of-the-art results on the UPTI dataset.

Scalable Approach To Construct Free-Standing and Flexible Carbon Networks for Lithium–Sulfur Battery

Reconstructing carbon nanomaterials (e.g., fullerene, carbon nanotubes (CNTs), and graphene) to multidimensional networks with hierarchical structure is a critical step in exploring their applications. Herein, a sacrificial template method by casting strategy is developed to prepare highly flexible and free-standing carbon film consisting of CNTs, graphene, or both. The scalable size, ultralight and binder-free characteristics, as well as the tunable process/property are promising for their large-scale applications, such as utilizing as interlayers in lithium-sulfur battery. The capability of holding polysulfides (i.e., suppressing the sulfur diffusion) for the networks made from CNTs, graphene, or their mixture is pronounced, among which CNTs are the best. The diffusion process of polysulfides can be visualized in a specially designed glass tube battery. X-ray photoelectron spectroscopy analysis of discharged electrodes was performed to characterize the species in electrodes. A detailed analysis of lithium diffusion constant, electrochemical impedance, and elementary distribution of sulfur in electrodes has been performed to further illustrate the differences of different carbon interlayers for Li-S batteries. The proposed simple and enlargeable production of carbon-based networks may facilitate their applications in battery industry even as a flexible cathode directly. The versatile and reconstructive strategy is extendable to prepare other flexible films and/or membranes for wider applications.

Multidimensional Approach to Online Interest Networks

Online social networks allow users to socialise with others and also construct and manage their identities via self-presentation. This creates a network of interests that is often represented by a single network that aggregates all of the connections, however it is possible to increase the possibilities by dividing it into layers of individual networks. In this paper we propose a model for a multidimensional network of interests based on the interests that online social network users provide on their profiles. This multidimensional approach can be useful for getting insights about user behaviour or even help organisations understand the impact of their online content.

Inferring Causal Direction From Multi-Dimensional Causal Networks for Assessing Harmful Factors in Security Analysis

In many scenarios of security analysis, Bayesian methods are often used for assessing harmful factors like attacks, as Bayesian causal model can estimate or predict the effect from the observed factors. To distinguish which ones are direct or significant causal (harmful) factors to the target systems, we often need to calculate a (local) causal network, but the existing methods usually return Markov equivalence classes instead of an actual causal structure. In this paper, a new approach for inferring causal direction from multi-dimensional causal networks for assessing harmful factors in security analysis is proposed based on a split-and-merge strategy. The method first decomposes an n-dimensional network into induced subnetworks, each of which corresponds to a node in the network. We show that each induced subnetwork can be subsumed into one of the three substructures: one-degree, non-triangle, and triangle-existence substructures. Three effective algorithms are developed to infer causalities from the three substructures. The whole causal structure of the multi-dimensional network is obtained by learning these induced subnetworks separately. Experimental results demonstrate that our method is more general and effective than the state-of-the-art methods.

The multidimensional role of science parks in attracting international knowledge migrants

There is a tendency to think in primarily economic terms (e.g., economic and financial opportunities) when considering how major science, technology and business spaces, also known as science parks (e.g., Silicon Valley, USA; Science Vale, UK), attract international knowledge migrants (IKMs) from the global knowledge economy. However, other elements that make places attractive for IKMs to achieve desirable standards of living are often not sufficiently addressed in the recent literature. Using a single case study of the Dutch science park Novel-T, this paper investigates how science parks can deliberately be created to attract and retain IKMs using a new multidimensional model for understanding the attractive effects of science parks on human capital. The fieldwork element involved 20 semi-structured face-to-face interviews, which were analyzed using a narrative analysis technique. The results of the case study show that science parks should be understood as multidimensional networks that trigger both a social attraction effect for IKMs as well as regional economic development and growth. The importance of the spatial design of science parks as well as the creation of a highly international atmosphere were found to be two influential factors in attracting IKMs to a particular science park in a peripheral region. Moreover, the results show that the deliberate construction of a science park with four intended operational effects is not a simple and controllable process, since some operational effects (e.g., external recognition) appear to be more controllable than others (e.g., internal social dynamics).

Multidimensional Data-Driven Life Prediction Method for White LEDs Based on BP-NN and Improved-Adaboost Algorithm

In view of the light-emitting diode (LED) and its life prediction, despite its currently wide use, IES TM-21-11 parametric life prediction method is incapable to extrapolate under multidimensional conditions (include working environmental conditions). This paper presents a multidimensional back propagation-neural network (BP-NN) based life prediction method which considers different driving currents and ambient temperatures. In this method, parameters such as temperature, electric current, initial chromaticity coordinates and initial luminous flux serve as the inputs, while life serves as the output. Since the traditional NN can easily get trapped in local minima and be affected by low precision, the BP-NN method is improved using Adaboost algorithm. The expected life predicted by the improved method is compared with that of traditional BP-NN and IES TM-21-11. The LED lamps of different power grades are compared for verification purposes. The results show that when predicting LED's lifetime, the improved method reduces the average relative error by on average by 54% compared with traditional BP-NN. However, the improved method takes 63.6% longer to operate, which requires the users to choose an appropriate model in accordance with particular operating conditions.

OPPORTUNITIES OF USE OF EXPERT KNOWLEDGE AS THE SOURCE OF CONCEPTS OF DEVELOPMENT OF THE ORGANIZATIONS

The purpose of article – to prove the approach expanding opportunities of use of expert knowledge as a source of formation of individual concepts of development of the organizations as social and economic objects. As is known, today there is no standard general theory of development. But there are individual theories or concepts of development which quantity has the expressed tendency of increase as in natural, and humanitarian areas. In these individual concepts development is understood differently, i.e. their authors suggest different forms of development. Development as an object of research is one of the most difficult and least investigated areas as in natural-science, and humanitarian sphere. In this connection it is necessary to ascertain, that, despite of the urgency, research of development in an insignificant degree provides practice by control facilities by development. Among experts there can be carriers of individual concepts of development, both expert-practice, and experts-theorists. Author’s concepts of experts are the most complete kind of knowledge containing advantages of a science, experience and practice. They contain features of rational analytical knowledge, intuitive-creative its irrational condition, and also feature of empirical sensually perceived knowledge – experience. Alongside with the quantitative approach in ideology of methods of the expert estimation basing on expert knowledge, there is also a qualitative approach. Thus, existing technology integration of individual experts concepts, as a representative of a qualitative approach, developed for the integration of individual experts and practitioners in the general concepts integrated concept. On a new level of the social and scientific importance the technology of integration of individual concepts in the complete general if the opportunity of integration of individual concepts of experts-theorists will be created can be lifted. But in this case other way, other requirements should be applied to integration of such individual concepts to formation of individual concepts of experts, their selection, to the form of a statement of the contents of concepts, their formal synthesis as a result of which the integrated complete concept is formed. Meeting the challenge of developing technology concepts integration of private experts-theorists in general theoretical concept, it is possible if you use an abstract theory of development, built on the basis of the corresponding construct. In a new approach to integrate the concepts of individual experts, theorists assumed the use of the methodology and conceptual analysis of the appropriate structural mathematics. As there is a theoretical groundwork article is proposed to use for the development of the theory of development as a multidimensional poliaspektnoy disjunctive network with increasing difficulty. For reasonable application of conceptual methods in solving the problem of the integration of the concepts of private experts, theorists need to consider the following features characterizing the state of the test domain. Firstly – it ourselves existing subject specific concepts of different authors to express their points of view on what’s inside the analyzed object, how it manifests itself, how the point of view of the author tested practice. All concepts should have the status of waste, of proven concepts with which he achieved a real effect. Typically, each of the concepts claim to universality, that it is able to explain the whole thing presented science (sociology, psychology, etc.), all of its phenomena. This effect is absolute expert of their private point of view. As among concepts can be such which with each other are not correlated, consequence of it is the aspiration of their authors to belittle other concepts, to declare them not to the full proved. Second – this is a feature that consists in the fact that science (or a separate domain), represented in partial concepts, general estimates of representatives of the science and their communities, just does not yet have a specific (“developed”) object, which allowed to conceptualize this area. The latter is represented by a mosaic weakly correlated with each other ideas. In theory and practice, the use of expert knowledge this statement of the problem is innovative. The problem is not developed. But it can be assumed that the concerted effort of sociologists and experts in the field of theoretical research tool – conceptual equipped with appropriate methods – will help to solve this difficult and urgent task.

Peptides Derived from Neuronal Cell Cycle like Kinase 5 Activator p35, in Neurodegeneration; Pathology and Therapy

Neurodegenerative disorders such as Alzheimer’s or Parkinson’s are diseases of aging. Mutations, lesions, neuronal insults, the genetic and stochastic “slings and arrows” of living, accumulate and downgrade organ function. Microarray analyses of aging brain identify patterns of genetic changes that correlate with neurodegenerative phenotypes. Genes involved in synaptic transmission, mitochondrial function and protein turnover are downregulated while significant numbers of genes regulating DNA repair, stress responses and inflammation are upregulated. This impact on the brain induces, a complex, multidimensional network of abnormal interactions leading to deposits of protein aggregates such as amyloid plaques, tau and neurofilament proteins tangles. For some, this pathology leads to neuron loss, behavioral defects, cognitive decline and death. Therapeutic approaches for AD, for example, target the molecular pathways leading to plaques and tangles; these include kinases and phosphatases. Among kinases, one stands out, Cdk5/p35, essential for neuronal migration, synapse formation, function and survival. Studies have shown that aging-induced neuronal stress deregulates and hyper activates Cdk5, a ubiquitous feature of neuronal disorders such as Alzheimer’s, Amyotropic lateral sclerosis (ALS) and Parkinson’s (PD). Among its many effects, hyperactive Cdk5 is implicated in the production of abnormal phosphorylated protein aggregates and is a therapeutic target. Roscovitine and related compounds inhibit Cdk5 activity but not specifically; cell cycle Cdks and other kinases are equally affected. In our laboratory two truncated peptides CIP (126a.a) and P5 (24a.a.), derived from p35, activator, of Cdk5 have been shown to specifically inhibit hyperactive Cdk5 in vitro, in cortical neurons and in AD, ALS and PD model mice. As a result the neurodegenerative phenotype was diminished; aggregates and inflammation were reduced, abnormal behavior was improved and increased animal’s longevity. We believe these peptides are excellent therapeutic candidates for those neurodegenerative disorders expressing hyperactive Cdk5 in the brain.

Multiview Community Discovery Algorithm via Nonnegative Factorization Matrix in Heterogeneous Networks

With the rapid development of the Internet and communication technologies, a large number of multimode or multidimensional networks widely emerge in real-world applications. Traditional community detection methods usually focus on homogeneous networks and simply treat different modes of nodes and connections in the same way, thus ignoring the inherent complexity and diversity of heterogeneous networks. It is challenging to effectively integrate the multiple modes of network information to discover the hidden community structure underlying heterogeneous interactions. In our work, a joint nonnegative matrix factorization (Joint-NMF) algorithm is proposed to discover the complex structure in heterogeneous networks. Our method transforms the heterogeneous dataset into a series of bipartite graphs correlated. Taking inspiration from the multiview method, we extend the semisupervised learning from single graph to several bipartite graphs with multiple views. In this way, it provides mutual information between different bipartite graphs to realize the collaborative learning of different classifiers, thus comprehensively considers the internal structure of all bipartite graphs, and makes all the classifiers tend to reach a consensus on the clustering results of the target-mode nodes. The experimental results show that Joint-NMF algorithm is efficient and well-behaved in real-world heterogeneous networks and can better explore the community structure of multimode nodes in heterogeneous networks.

CcNET: Database of co-expression networks with functional modules for diploid and polyploid Gossypium.

Plant genera with both diploid and polyploid species are a common evolutionary occurrence. Polyploids, especially allopolyploids such as cotton and wheat, are a great model system for heterosis research. Here, we have integrated genome sequences and transcriptome data of Gossypium species to construct co-expression networks and identified functional modules from different cotton species, including 1155 and 1884 modules in G. arboreum and G. hirsutum, respectively. We overlayed the gene expression results onto the co-expression network. We further provided network comparison analysis for orthologous genes across the diploid and allotetraploid Gossypium We also constructed miRNA-target networks and predicted PPI networks for both cotton species. Furthermore, we integrated in-house ChIP-seq data of histone modification (H3K4me3) together with cis-element analysis and gene sets enrichment analysis tools for studying possible gene regulatory mechanism in Gossypium species. Finally, we have constructed an online ccNET database (http://structuralbiology.cau.edu.cn/gossypium) for comparative gene functional analyses at a multi-dimensional network and epigenomic level across diploid and polyploid Gossypium species. The ccNET database will be beneficial for community to yield novel insights into gene/module functions during cotton development and stress response, and might be useful for studying conservation and diversity in other polyploid plants, such as T. aestivum and Brassica napus.

Multidimensional wavelet neural networks Based on polynomial powers of sigmoid

Wavelet functions have been used as the activation function in feed forward neural networks. An abundance of R&D has been produced on wavelet neural network area. Some successful algorithms and applications in wavelet neural network have been developed and reported in the literature. However, most of the aforementioned reports impose many restrictions in the classical back propagation algorithm, such as low dimensionality, tensor product of wavelets, parameters initialization, and, in general, the output is one dimensional, etc. In order to remove some of these restrictions, a family of polynomial wavelets generated from powers of sigmoid functions is presented. We described how a multidimensional wavelet neural networks based on these functions can be constructed, trained and applied in pattern recognition tasks. As examples of applications for the method proposed a framework for face verfication is presented.

Reconstruction of multidimensional carbon hosts with combined 0D, 1D and 2D networks for enhanced lithium-sulfur batteries

Abstract It is a core task to find solutions to suppress the “shuttle effect” of polysulfides and improve high rate capability at the sulfur cathode of lithium sulfur batteries. Herein we first time propose a concept of multileveled blocking “dams” to suppress the diffusion of polysulfides. We report a facile and effective strategy to construct multidimensional conductive carbon hosts for accommodation of active sulfur. Multidimensional ternary carbon networks (MTCNs) with 0D nanospheres, 1D nanotubes and 2D nanoflakes are organically combined together to provide multileveled conductive channels to reserve active sulfur and promote stable sustained reactions. In the light of enhanced conductivity and multileveled blocking “dams” for polysulfides, the designed MTCNs/S cathode has been demonstrated with noticeable improvement in discharge capacity (1472 mAh g −1 at 0.l C) and long-term cycling stability (65% retention at 5.0 C after 500 cycles). Our research may provide a new insight in the gradient blocking of polysulfides with the help of multidimensional carbon networks.

Integrating Multidimensional Data Sources to Identify Genes Regulating Complex Phenotypes.

Phenotypes collected with a view to quantitative trait locus mapping can be augmented with compatible whole-transcriptome expression data and information from several other sources. These different data sources can be assembled into multidimensional network models which allow the identification of key genes potentially driving the phenotype of interest. The following chapter describes this approach using an example workflow. Several alternatives and potential limitations are discussed to aid the researcher when applying these techniques to their own work.

Co-expression network analyses identify functional modules associated with development and stress response in Gossypium arboreum.

Cotton is an economically important crop, essential for the agriculture and textile industries. Through integrating transcriptomic data, we discovered that multi-dimensional co-expression network analysis was powerful for predicting cotton gene functions and functional modules. Here, the recently available transcriptomic data on Gossypium arboreum, including data on multiple growth stages of tissues and stress treatment samples were applied to construct a co-expression network exploring multi-dimensional expression (development and stress) through multi-layered approaches. Based on differential gene expression and network analysis, a fibre development regulatory module of the gene GaKNL1 was found to regulate the second cell wall through repressing the activity of REVOLUTA, and a tissue-selective module of GaJAZ1a was examined in response to water stress. Moreover, comparative genomics analysis of the JAZ1-related regulatory module revealed high conservation across plant species. In addition, 1155 functional modules were identified through integrating the co-expression network, module classification and function enrichment tools, which cover functions such as metabolism, stress responses, and transcriptional regulation. In the end, an online platform was built for network analysis (http://structuralbiology.cau.edu.cn/arboreum), which could help to refine the annotation of cotton gene function and establish a data mining system to identify functional genes or modules with important agronomic traits.

Boundary-spanning Knowledge Integration of “Internet Plus Manufacturing”: Based on the Governance of Multi-Level and Multi-Dimensional Knowledge Network

Under the context of economic transition, exploring a new format of "Internet plus Manufacturing" which was based on the emerging information technology platform and the powerful manufacturing industry basis in Zhejiang Province, has an enormous effect in promoting the transforming and upgrading of manufacturing industry. From the perspective of knowledge network governance, this paper reveals the theory of boundary-spanning knowledge integration of "Internet Plus manufacturing" enterprises, which is of great theoretical significance to reconstruction of the existing theory. The conclusion will provide the theoretical basis of the knowledge network management and boundary-spanning knowledge integration for the Internet Plus manufacturing firms and enlighten policy decision about "Internet Plus manufacturing".

Multi-dimensional Network Security Situation Assessment

Multi-dimensional Network Security Situation Assessment

Urdu Nasta'liq text recognition using implicit segmentation based on multi-dimensional long short term memory neural networks.

The recognition of Arabic script and its derivatives such as Urdu, Persian, Pashto etc. is a difficult task due to complexity of this script. Particularly, Urdu text recognition is more difficult due to its Nasta’liq writing style. Nasta’liq writing style inherits complex calligraphic nature, which presents major issues to recognition of Urdu text owing to diagonality in writing, high cursiveness, context sensitivity and overlapping of characters. Therefore, the work done for recognition of Arabic script cannot be directly applied to Urdu recognition. We present Multi-dimensional Long Short Term Memory (MDLSTM) Recurrent Neural Networks with an output layer designed for sequence labeling for recognition of printed Urdu text-lines written in the Nasta’liq writing style. Experiments show that MDLSTM attained a recognition accuracy of 98% for the unconstrained Urdu Nasta’liq printed text, which significantly outperforms the state-of-the-art techniques.

CcNET: Database of co-expression networks with functional modules for diploid and polyploid Gossypium.

Plant genera with both diploid and polyploid species are a common evolutionary occurrence. Polyploids, especially allopolyploids such as cotton and wheat, are a great model system for heterosis research. Here, we have integrated genome sequences and transcriptome data of Gossypium species to construct co-expression networks and identified functional modules from different cotton species, including 1155 and 1884 modules in G. arboreum and G. hirsutum, respectively. We overlayed the gene expression results onto the co-expression network. We further provided network comparison analysis for orthologous genes across the diploid and allotetraploid Gossypium. We also constructed miRNA-target networks and predicted PPI networks for both cotton species. Furthermore, we integrated in-house ChIP-seq data of histone modification (H3K4me3) together with cis-element analysis and gene sets enrichment analysis tools for studying possible gene regulatory mechanism in Gossypium species. Finally, we have constructed an online ccNET database (http://structuralbiology.cau.edu.cn/gossypium) for comparative gene functional analyses at a multi-dimensional network and epigenomic level across diploid and polyploid Gossypium species. The ccNET database will be beneficial for community to yield novel insights into gene/module functions during cotton development and stress response, and might be useful for studying conservation and diversity in other polyploid plants, such as T. aestivum and Brassica napus.

Logic Gate-based Evolutionary Algorithm for the multidimensional knapsack problem-wireless sensor network application

Evolutionary algorithms (EAs) are predominantly employed to find solutions for continuous optimization problems. As EAs are initially presented for continuous spaces, research on extending EAs to find solutions for binary spaces is in growing concern. In this paper, a logic gate-based evolutionary algorithm (LGEA) for solving some combinatorial optimization problems (COPs) is introduced. The proposed LGEA has the following features. First, it employs the logic operation to generate the trial population. Thereby, LGEA replaces common space transformation rules and classic recombination and mutation methods. Second, it is based on exploiting a variety of logic gates to search for the best solution. The variety among these logic tools will naturally lead to promote diversity in the population and improve global search abilities. The LGEA presents thus a new technique to combine the logic gates into the procedure of generating offspring in an evolutionary context. To judge the performance of the algorithm, we have solved the NP-hard multidimensional knapsack problem as well as a well-known engineering optimization problem, task allocation for wireless sensor network. Experimental results show that the proposed LGEA is promising.

Hardware-programmable optical networks

For future multi-dimensional optical networks, vast network resources provided by space division multiplexing and wavelength division multiplexing technologies, require new network architectures to scale up current network functions. The huge switch-granularity range requires a more dynamic way to deploy network resources. In this paper, we proposed a hardware-programmable optical network which deploys network resources according to incoming traffic requests. The proposed network supports node function programmability and node architecture adaptability, which are critical for dynamic function and resources deployments. Architecture-on-Demand based node architecture adapts node architectures and also enables network function programmability by incorporating with several flexible node functions. Other enabling technologies, such as ubiquitous power monitoring and dynamic optical power management, assures the programmable optical node work properly. Based on all these technologies, we established a hardware-programmable optical network testbed. Several use cases were demonstrated successfully, such as dynamic power equalization and optical debugging. These work verified the feasibility of hardware-programmable optical network, which dynamically allocate network resources for service provision. The proposed hardware programmable optical network will lead to a better hardware utilization and provide a possible solution for the future multi-dimensional optical network.