Dimensional Network Research Articles

Defect recognition of photovoltaic panels based on adaptive dimensional feature aggregation convolutional neural network

Abstract Under the policy framework of achieving carbon neutrality goals and addressing climate change, the importance of photovoltaic power generation will become more prominent, therefore the detection of defects in photovoltaic panels will become increasingly important. In order to accurately detect defects in photovoltaic panels, this paper proposes an adaptive dimensional feature aggregation algorithm that combines Deformable Convolutional (DCN) and C3 convolutional layers to form C3_ DCN module enhances the generalization ability of convolutional kernels; By incorporating the Cross Modal Transformer Attention Mechanism (CoTAttention) and Context Enhancement Module (CAM) at the output end, the model has the ability to adaptively allocate computing resources and enhance the ability to detect small targets; Using Focal EIoU as the loss function reduces the slow convergence speed and inaccurate results of known box regression methods. The experimental results show that for the PVEL-AD industrial detection dataset, the improved model has an average accuracy of 91.8%, which is 1.5% higher than the original model.

An ensemble deep learning-based cyber attack detection system using optimization strategy

Background of the studyCyber attack stories become a routine in which new levels of intention are shown by the cyber attacker through sophisticated attacks on networks. The conventional approaches struggle to detect unknown malware and complex attacks. It does not secure the user's individual information. Hence, deep learning methods are used to effectively predict the attacks. Proposed modelThis paper implements an intelligent cyber security mechanism by employing effective mechanisms. Initially, the significant data for this approach is gathered from the benchmark data sources, and then they are given into the optimal feature chosen. Here, the parameters are optimally chosen with the aid of the Modified Puzzle Optimization Algorithm (MPOA) for the successive outcomes. Moreover, the optimally selected attributes are fed into the Ensemble Serial Cascaded Deep learning with Attention Mechanism (ESCDLAM), which is the integration of the 1 Dimensional Convolutional Neural Networks (1DCNN), Recurrent Neural Network (RNN), and Deep Temporal Convolutional Networks (DTCN). Here, the parameter optimization takes place to improve detection performance. DatasetThe experimentation is conducted through a diverse datasets which includes CICIDS 2017 Friday Noon DDoS 0beefa93–4, NSL-KDD-Dataset, and Kddcup99 for the effective experimental analysis for detecting the cyber attacks. Findings of the resultsThe results outcome of the developed model shows 97 % in terms of accuracy, recall, MCC, and specificity. Also, the precision of the developed model attains a value of 95. ConclusionThe outcomes of the approach are compared with the other pre-existing approaches to show its supremacy.

Application of Network Dimension Theory to the Kinetics of Nanogel Formation in Miniemulsion Vinyl/Divinyl Copolymerization: Free‐Radical and Living Polymerization

AbstractIn vinyl/divinyl copolymerization, a crosslink is formed by the reaction between an active center and a pendant double bond. When both the active center and the pendant double bond are located within the same polymer molecule, the cyclization occurs, which is ineffective for growth in molecular weight. In the present model, the network dimension theory is applied to estimate the mean‐square radius of gyration for the growing polymer molecule, which is used to account for the enrichment effect of pendant double bonds around the active center for the cyclization reaction. The model is applied to the miniemulsion copolymerization, and both conventional free‐radical polymerization and ideal living polymerization are considered. Some of important characteristics of network architecture formed in these two types of polymerization mechanisms that cannot be predicted based on the classical chemical kinetics can be reproduced by the model, such unique characteristics as the pendant double bonds are consumed from the beginning of polymerization in the conventional free‐radical polymerization but not so in the living polymerization. The present model provides useful insights into the size and structural dependent network formation kinetics without relying on the lattice model.

Networks Cultivating Values: Insights from Five Culture-based Regeneration Projects in Italy

This contribution aims at improving the understanding of the conditions and dynamics that facilitate processes of culture-based urban regeneration via social innovation. It presents and discusses the outcomes of an in-depth investigation of five projects taking place in large cities in Italy dealing with the reuse and regeneration of urban spaces through creative and innovative practices. Research findings demonstrate the centrality of the network dimension in sustaining and shaping processes of urban regeneration through social innovation and cultural production. Networks are led by social entrepreneurs having a strong territorial focus that “cultivate” new values for the local community by reusing and mobilizing publicly owned vacant or abandoned material assets.

Impact of Personal Networks of Ethnic Minority Female Entrepreneurs on Growth Construct: Blend of Egocentric and Narrative Analysis

The paper aims to explore and explain the content and structure of personal networks of immigrant and British-born Pakistani female entrepreneurs for business growth. Pakistani female entrepreneurs represent a distinct segment of ethnic minority entrepreneurs as they represent an interesting blend of socio-economic, religious and socio-cultural traits. This research study underscores the interactional dimension of personal networks and its impact on the growth of their business. It is mainly the contents and interactions within a network that determines the quality of a network and its influence on relevant aspects of network members. In order to explore the meaning and perceptions attached to relationships and their use for entrepreneurial ventures’ growth, an egocentric network analysis is applied to narratives of Pakistani female entrepreneurs (the ego). Narratives were collected through in-depth interviews with ‘name generator’ questions. Egocentric network analysis was carried out using sociograms to explore the formation and nature of network ties. By generating an in depth understanding of the distinctive use of personal networks for growth of business by Pakistani female entrepreneurs the paper provides in depth knowledge and understanding of a particular ethnic minority group that will be of use to business owners (Pakistani female entrepreneurs), academic researchers and policy makers.

Viscous Elastic Asphalt Characteristics of Cariphalte Modification with Addition of Gilsonite Viewed from Phase Angle Value

Performance Grade (PG) Bitumen is bitumen that is graded based on its performance at different temperatures. In the Superpave grading system, asphalt binders are classified according to their performance in extreme hot and cold temperatures and are referred to as performance grade (PG) asphalt. The main objective of assessing and selecting asphalt binder using the PG system is to ensure that the binder has properties suitable for the environmental conditions in the field. Cariphalte is a premium grade polymer modified bitumen used in road and airport construction and other similar applications. Cariphalte is a Styrene Butadine Styrene (SBS) modified bitumen. The SBS in Cariphalte ensures the formation of a 3-dimensional network in the bitumen thereby reducing the temperature susceptibility and increasing the stiffness modulus at high temperatures. Cariphalte is also resistant to oxidation, reducing deformation and slowing secondary solidification. To get an idea of the improvement in rheological properties due to the addition of gilsonite, tests were carried out on a mixture of cariphalte and gilsonite asphalt with 6 variations of gilsonite content. The tests carried out include basic rheological properties testing and mechanistic rheological properties testing. From this research, the test results showed that at the same temperature, asphalt with the addition of 0% gilsonite had the smallest phase angle (δ) compared to asphalt using the addition of gilsonite. This explains that the addition of gilsonite can increase the phase angle (δ) value of PG-76 asphalt, which makes the asphalt material have thicker characteristics.

ImageDTA: A Simple Model for Drug-Target Binding Affinity Prediction.

Predicting the drug-target binding affinity (DTA) is crucial in drug discovery, and an increasing number of researchers are using artificial intelligence techniques to make such predictions. Many effective deep neural network prediction models have been proposed. However, current methods need improvement in accuracy, complexity, and efficiency. In this study, we propose a method based on a multiscale 2-dimensional convolutional neural network (CNN), namely ImageDTA. Many studies have shown that CNN achieves good learning effects with limited data. Therefore, we take a unique perspective by treating the word vector encoded with a simplified molecular input line entry system (SMILES) string as an "image" and processing it like handling images, fully leveraging the efficient processing capabilities of CNN for image data. Furthermore, we show that ImageDTA has higher training and inference efficiency than pretrained large models and outperforms attention-based graph neural network models in accuracy and interpretability. We also use visualization techniques to select appropriate convolutional kernel sizes, thereby increasing the network's interpretability.

Robust Structure-Aware Graph-based Semi-Supervised Learning: Batch and Recursive Processing

Graph-based semi-supervised learning plays an important role in large scale image classification tasks. However, the problem becomes very challenging in the presence of noisy labels and outliers. Moreover, traditional robust semi-supervised learning solutions suffers from prohibitive computational burdens thus cannot be computed for streaming data. Motivated by that, we present a novel unified framework robust structure-aware semi-supervised learning called Unified RSSL (URSSL) for batch processing and recursive processing robust to both outliers and noisy labels. Particularly, URSSL applies joint semi-supervised dimensionality reduction with robust estimators and network sparse regularization simultaneously on the graph Laplacian matrix iteratively to preserve the intrinsic graph structure and ensure robustness to the compound noise. First, in order to relieve the influence from outliers, a novel semi-supervised robust dimensionality reduction is applied relying on robust estimators to suppress outliers. Meanwhile, to tackle noisy labels, the denoised graph similarity information is encoded into the network regularization. Moreover, by identifying strong relevance of dimensionality reduction and network regularization in the context of robust semi-supervised learning (RSSL), a two-step alternative optimization is derived to compute optimal solutions with guaranteed convergence. We further derive our framework to adapt to large scale semi-supervised learning particularly suitable for large scale image classification and demonstrate the model robustness under different adversarial attacks. For recursive processing, we rely on reparameterization to transform the formulation to unlock the challenging problem of robust streaming-based semi-supervised learning. Last but not least, we extend our solution into distributed solutions to resolve the challenging issue of distributed robust semi-supervised learning when images are captured by multiple cameras at different locations. Extensive experimental results demonstrate the promising performance of this framework when applied to multiple benchmark datasets with respect to state-of-the-art approaches for important applications in the areas of image classification and spam data analysis.

Remaining useful life prediction of machinery based on improved Sample Convolution and Interaction Network

Remaining useful life (RUL) prediction is significant in ensuring the safe and reliable operation of machinery and reducing maintenance costs. There are currently numerous deep learning-based methods for machinery RUL prediction. However, some studies overlook the differences in the contributions of data from different sensors or different time points of the same sensor, and most research only extracts information from feature or sequence dimensions, which inevitably affects the efficiency and accuracy of RUL prediction. Therefore, we proposed a method based on a multi-dimensional attention mechanism and feature-sequence dimensional sample convolution and interaction network (MFSSCINet) to predict the machinery RUL effectively, which includes a Feature-Sequence Dimension Attention Module to capture information interactions in feature dimension and learn the impact weights of various time steps in sequence dimension. Then, a Multi-Source Information Fusion Module was constructed to extract helpful information from features of different dimensions and time resolutions and fuse them. Finally, a RUL Prediction Module was built to estimate the machine RUL effectively. The method’s effectiveness is validated in C-MAPSS and XJTU-SY datasets. Experimental results show that the MFSSCINet model has higher accuracy in machine RUL prediction tasks than other advanced computational methods.

A Mesoporous Gold Sensor Unveils Phospho PD-L1 in Extracellular Vesicles as a Proxy for PD-L1 Expression in Lung Cancer Tissue.

Immune checkpoint inhibitors (ICIs) targeting programmed cell death ligand 1 (PD-L1), or its receptor, PD-1 have improved survival in patients with non-small-cell lung cancer (NSCLC). Assessment of PD-L1 expression requires tissue biopsy or fine needle aspiration that are currently used to identify patients most likely to respond to single agent anti-PD-1/PD-L1 therapy. However, obtaining sufficient tissue to generate a PD-L1 tissue proportion score (TPS) ≥ 50% using immunohistochemistry remains a challenge that potentially may be overcome by liquid biopsies. This study utilized a mesoporous gold sensor (MGS) assay to examine the phosphorylation status of PD-L1 in plasma extracellular vesicles (EV pPD-L1) and PD-L1 levels in plasma from NSCLC patient samples and their association with tumor PD-L1 TPS. The 3-dimensional mesoporous network of the electrodes provides a large surface area, high signal-to-noise ratio, and a superior electro-conductive framework, thereby significantly improving the detection sensitivity of PD-L1 nanosensing. Test (n = 20) (Pearson's r = 0.99) and validation (n = 45) (Pearson's r = 0.99) cohorts show that EV pPD-L1 status correlates linearly with the tumor PD-L1 TPS assessed by immunohistochemistry irrespective of the tumor stage, with 64% of patients overall showing detectable EV pPD-L1 levels in plasma. In contrast to the EV pPD-L1 results, plasma PD-L1 levels did not correlate with the tumor PD-L1 TPS score or EV pPD-L1 levels. These data demonstrate that EV pPD-L1 levels may be used to select patients for appropriate PD-1 and PD-L1 ICI therapy regimens in early, locally advanced, and advanced NSCLC and should be tested further in randomized controlled trials. Most importantly, the assay used has a less than 24h turnaround time, facilitating adoption of the test into the routine diagnostic evaluation of patients prior to therapy.

Connectome embedding in multidimensional graph spaces

Abstract Connectomes' topological organization can be quantified using graph theory. Here, we investigated brain networks in higher dimensional spaces defined by up to ten graph-theoretic nodal properties. These properties assign a score to nodes, reflecting their meaning in the network. Using 100 healthy unrelated subjects from the Human Connectome Project, we generated various connectomes (structural/functional, binary/weighted). We observed that nodal properties are correlated (i.e., they carry similar information) at whole-brain and subnetwork level. We conducted an exploratory machine learning analysis to test whether high dimensional network information differs between sensory and association areas. Brain regions of sensory and association networks were classified with an 80-86% accuracy in a 10D space. We observed the largest gain in machine learning accuracy going from a 2D to 3D space, with a plateauing accuracy towards 10D space, and nonlinear Gaussian kernels outperformed linear kernels. Finally, we quantified the Euclidean distance between nodes in a 10D graph space. The multidimensional Euclidean distance was highest across subjects in the default mode network (in structural networks) and frontoparietal and temporal lobe areas (in functional networks). To conclude, we propose a new framework for quantifying network features in high-dimensional spaces that may reveal new network properties of the brain.

Network Spreading from Network Dimension.

Continuous-state network spreading models provide critical numerical and analytic insights into transmission processes in epidemiology, rumor propagation, knowledge dissemination, and many other areas. Most of these models reflect only local features such as adjacency, degree, and transitivity, so can exhibit substantial error in the presence of global correlations typical of empirical networks. Here, we propose mitigating this limitation via a network property ideally suited to capturing spreading. This is the network correlation dimension, which characterizes how the number of nodes within range of a source typically scales with distance. Applying the approach to susceptible-infected-recovered processes leads to a spreading model which, for a wide range of networks and epidemic parameters, can provide more accurate predictions of the early stages of a spreading process than important established models of substantially higher complexity. In addition, the proposed model leads to a basic reproduction number that provides information about the final state not available from popular established models.

High-accuracy models for iris recognition with merging features

Due to advancements in computer science and technology, impersonation has become more common. Today, biometrics technology is widely used in various aspects of people's lives. Iris recognition, known for its high accuracy and speed, is a significant and challenging field of study. As a result, iris recognition technology and biometric systems are utilized for security in numerous applications, including human-computer interaction and surveillance systems. It is crucial to develop advanced models to combat impersonation crimes. This study proposes sophisticated artificial intelligence models with high accuracy and speed to eliminate these crimes. The models use linear discriminant analysis (LDA) for feature extraction and mutual information (MI), along with analysis of variance (ANOVA) for feature selection. Two iris classification systems were developed: one using LDA as an input for the OneR machine learning algorithm and another innovative hybrid model based on a One Dimensional Convolutional Neural Network (HM-1DCNN). The MMU database was employed, achieving a performance measure of 94.387% accuracy for the OneR model. Additionally, the HM-1DCNN model achieved 99.9% accuracy by integrating LDA with MI and ANOVA. Comparisons with previous studies show that the HM-1DCNN model performs exceptionally well, with at least 1.69% higher accuracy and lower processing time.

Kekuatan Modal Sosial Pada Kelompok Tani Jagung Di Desa Suwaan Kecamatan Kalawat Kabupaten Minahasa Utara

This research aims to describe the social capital strengths of the Anugrah Corn Farmer Group in Suwaan Village, Kalawat District, North Minahasa Regency. The study took place over 3 months from August to October 2023. It was conducted within the corn farmer group in Suwaan Village, Kalawat District, North Minahasa Regency. Data for this research were collected using primary sources through direct interviews with village elders/secretaries and community members of the farmer group, supported by observations and questionnaires as data collection tools. Secondary data were obtained from relevant agencies such as the village office and the internet to supplement primary data. Purposive sampling was employed to select participants, specifically the Anugrah Corn Farmer Group in Suwaan Village, totaling 11 farmers. Data analysis utilized descriptive analysis with Likert scale measurements using tables and figures. The research findings indicate that social capital played a significant role: the total trust dimension score was 171, categorized as very high; the total norm dimension score was 153, also categorized as very high; and the network dimension yielded an overall score of 492 from 11 respondents of the Anugrah Corn Farmer Group, representing 93.18%, categorized as very high.

Computing edge version of metric dimension of certain chemical networks

In the modern digital sphere, graph theory is a significant field of research that has a great deal of significance. It finds widespread application in computer science, robotic directions, and chemistry. Additionally, graph theory is used in robot network localization, computer network problems and the formation of various chemical structures for networks. Moreover, it finds uses in exploring diffusion mechanisms and scheduling aircraft as well. The present research project examines and concentrates on the edge version of metric dimension of the Concealed Non-Kekuléan Benzenoid Hydrocarbon, Polythiophene, Backbone DNA network and Bakelite networks. All the mentioned networks have constant edge metric dimension except Bakelite network, as demonstrated by the results. If we talk about the applications of these networks, Polythiophene are used to treat prion disorders. It is also capable of detecting metal ions. The concept of Bakelite, which finds applications in the jewelry, electrical, cookware, sports, and clock industries, had an impact on the invention of modern polymers. The functions of DNA include information encoding, replication, mutation, and recombination gene expression.

3DVascNet: An Automated Software for Segmentation and Quantification of Mouse Vascular Networks in 3D.

Analysis of vascular networks is an essential step to unravel the mechanisms regulating the physiological and pathological organization of blood vessels. So far, most of the analyses are performed using 2-dimensional projections of 3-dimensional (3D) networks, a strategy that has several obvious shortcomings. For instance, it does not capture the true geometry of the vasculature and generates artifacts on vessel connectivity. These limitations are accepted in the field because manual analysis of 3D vascular networks is a laborious and complex process that is often prohibitive for large volumes. To overcome these issues, we developed 3DVascNet, a deep learning-based software for automated segmentation and quantification of 3D retinal vascular networks. 3DVascNet performs segmentation based on a deep learning model, and it quantifies vascular morphometric parameters such as vessel density, branch length, vessel radius, and branching point density. We tested the performance of 3DVascNet using a large data set of 3D microscopy images of mouse retinal blood vessels. We demonstrated that 3DVascNet efficiently segments vascular networks in 3D and that vascular morphometric parameters capture phenotypes detected by using manual segmentation and quantification in 2 dimension. In addition, we showed that, despite being trained on retinal images, 3DVascNet has high generalization capability and successfully segments images originating from other data sets and organs. Overall, we present 3DVascNet, a freely available software that includes a user-friendly graphical interface for researchers with no programming experience, which will greatly facilitate the ability to study vascular networks in 3D in health and disease. Moreover, the source code of 3DVascNet is publicly available, thus it can be easily extended for the analysis of other 3D vascular networks by other users.

Serial cascaded deep feature extraction-based adaptive attention dilated model for crop recommendation framework

Effective crop farming depends on wise selection of crops. It is an essential factor that has to be fulfilled before beginning an agricultural endeavor. Conventionally, the crop that has to be grown is selected without considering the location and cultivated site’s characteristics by only considering its profit and demand on the market. Choosing the best crop for the circumstances can minimize the need for additional fertilizer and water for irrigation and help in attaining enhanced crop yield. Therefore, choosing the right crop is crucial for a successful agricultural situation. Thus, a novel crop recommendation model by considering the soil and geographical conditions is developed to aid the farmers in choosing the appropriate crop for the right condition so that the overall production can be enhanced to increase the overall profit and decrease the losses faced by the farmers. At first, a certain geographical area is selected, and the ideal parameters for growing a particular plant are gathered from the standard database. Next, the deep optimal features are extracted using a Serial Cascaded network in which an autoencoder is cascaded with a “Dimensional Convolutional Neural Network (1DCNN)” from the gathered data. The obtained deep features are optimally selected using the developed Modified Movement Territory of Fire Hawk Optimizer (MMTFHO). These optimally selected features are given to the Adaptive and Attention-based Hybrid Network (AAHNet) in which “Gated Recurrent Unit (GRU), and Long Short Term Memory (LSTM)” are utilized for choosing the right crop for the provided geographical condition. The parameters in the AAHNet are optimized using the same enhanced MMTFHO algorithm for improving the precision of the appropriate crop selection process. The final prediction of crops for the given geographical condition is obtained from the AAHNet. The final or overall rating of the recommended approach regarding accuracy metrics is 96.73%.

Finite dimension of the ion pathway networks in conducting glasses

In disordered materials, the ordinary understanding is that charge carriers tend to occupy energetically favorable sites known as ion-conducting channels. Many studies have revealed that the inherent fractal properties of such pathways lead to a sub-diffusive behavior. The linearity or branching of these pathways is crucial for determining how the charge carriers move. It can be thought that as the space dimensionality decreases, the average distance between the highest energy barriers along the conduction paths increases. In this study the finite dimension of those pathways is computed using an extended version of the classical Hausdorff dimension. Also, the Arrhenius behavior of the most mobile lithium ions is proved, confirming that such are responsible for conductivity behavior. The lithium ions mobility behavior in response to temperature changes and the finite dimension allowed to identify the ion diffusion regions fractal features. The reported results demonstrate that as the temperature increases the conducting channels become broadener, facilitating the transfer of electrical charge through the glassy matrix, below the transition temperature. The pathways behavior confirms the increase of the ionic conductivity when the temperature increases as it is experimentally observed.

Feature selection for high-dimensional neural network potentials with the adaptive group lasso

Neural network potentials are a powerful tool for atomistic simulations, allowing to accurately reproduce ab initio potential energy surfaces with computational performance approaching classical force fields. A central component of such potentials is the transformation of atomic positions into a set of atomic features in a most efficient and informative way. In this work, a feature selection method is introduced for high dimensional neural network potentials, based on the adaptive group lasso (AGL) approach. It is shown that the use of an embedded method, taking into account the interplay between features and their action in the estimator, is necessary to optimize the number of features. The method’s efficiency is tested on three different monoatomic systems, including Lennard–Jones as a simple test case, Aluminium as a system characterized by predominantly radial interactions, and Boron as representative of a system with strongly directional components in the interactions. The AGL is compared with unsupervised filter methods and found to perform consistently better in reducing the number of features needed to reproduce the reference simulation data at a similar level of accuracy as the starting feature set. In particular, our results show the importance of taking into account model predictions in feature selection for interatomic potentials.

Bulk transparent supramolecular glass enabled by host–guest molecular recognition

Supramolecular glass is a non-covalently cross-linked amorphous material that exhibits excellent optical properties and unique intrinsic structural features. Compared with artificial inorganic/organic glass, which has been extensively developed, supramolecular glass is still in the infancy stage, and itself is rarely recognized and studied thus far. Herein, we present the development of the host–guest molecular recognition motifs between methyl-β-cyclodextrin and para-hydroxybenzoic acid as the building blocks of supramolecular glass. Non-covalent polymerization resulting from the host–guest complexation and hydrogen bonding formation enables high transparency and bulk state to supramolecular glass. Various advantages, including recyclability, compatibility, and thermal processability, are associated with dynamic assembly pattern. Short-range order (host–guest complexation) and long-range disorder (three dimensional polymeric network) structures are identified simultaneously, thus demonstrating the typical structural characteristics of glass. This work provides a supramolecular strategy for constructing transparent materials from organic components.