Random Network Model Research Articles

Dimension of Activity in Random Neural Networks.

Neural networks are high-dimensional nonlinear dynamical systems that process information through the coordinated activity of many connected units. Understanding how biological and machine-learning networks function and learn requires knowledge of the structure of this coordinated activity, information contained, for example, in cross covariances between units. Self-consistent dynamical mean field theory (DMFT) has elucidated several features of random neural networks-in particular, that they can generate chaotic activity-however, a calculation of cross covariances using this approach has not been provided. Here, we calculate cross covariances self-consistently via a two-site cavity DMFT. We use this theory to probe spatiotemporal features of activity coordination in a classic random-network model with independent and identically distributed (i.i.d.) couplings, showing an extensive but fractionally low effective dimension of activity and a long population-level timescale. Our formulas apply to a wide range of single-unit dynamics and generalize to non-i.i.d. couplings. As an example of the latter, we analyze the case of partially symmetric couplings.

Magnetic‐optical dual functional Janus particles for the detection of metal ions assisted by machine learning

AbstractFunctional polymer microspheres have broad application prospects in various fields, such as metal ion detection, adsorption, separation, and controlled drug release. However, integrating different functions in a single microsphere system is a significant challenge in this field. In this work, we prepared multicompartmental emulsion droplets utilizing microfluidic technology. Fe3O4 magnetic nanoparticles were added to one of the compartments of the emulsion droplets as functional particles, and Janus microspheres were obtained after curing. Fluorescent probes enter the two compartments of the Janus microspheres by diffusion. The fluorescence changes of the microspheres were observed in situ and captured through a fluorescence microscope. The images are processed by image recognition software and a Python program. The “fingerprint” of the detected metal ions is obtained by dimensionality reduction of the data through Principal Component Analysis. We employ different algorithms to build Machine Learning models for predicting the metal ion species and concentration. The variation of fluorescence intensity of the three fluorescent probes and the corresponding R, G, and B channel values and time are used as descriptors. The results show that the Random Forest, K‐neighborhood (KNN), and Neural Network models demonstrated a better predicted effect with a variance (R2) greater than 0.9 and a smaller root mean square error; among them, the KNN model predicted the most accurate results.

Coupled Analysis of Safety Risks in Bridge Construction Based on N-K Model and SNA

Bridge construction accidents are often caused by a variety of factors, so it is particularly important to explore the role mechanism of bridge construction accident risk factors to effectively prevent construction safety accidents and ensure the smooth construction of bridges. We collect the causes of bridge construction accidents in China from 2006 to 2023, take 126 typical cases as research samples, analyze the primary risk factors of bridge construction from four aspects (human factors, equipment factors, management factors, and environmental factors), establish a library of secondary risk factors with reference to the literature research, introduce the theory of risk coupling, and analyze the coupling mechanism and types of risk factors of bridge construction accidents. The N-K random Boolean network model (N-K model) quantifies the coupling relationship between risk factors, assesses the risk level, and uses social network analysis (SNA) to analyze the network of bridge construction accident risk factors. The results indicate that the more factors involved in risk coupling, the greater the safety risks in bridge construction. Human factors are susceptible to the influence of other elements, and environmental and management factors can directly or indirectly impact other factors. In addition, operational errors, a lack of supervision and management, inadequate safety inspections, poor management personnel, and insufficient technical capabilities are also key risk factors that need to be prevented and controlled.

Effect of preparation conditions on the structure of As-Se glasses: Reverse Monte Carlo simulation

As an extension to part I, Dongol and Elhady (2020), the method of Reverse Monte Carlo simulation has been used to examine the effect of quenching rate on the structure of glassy AsxSe100−x (x=20, 30, 40, 50 at%). Results of the partial structure factors indicate that the correlations responsible for medium range order depend strongly on As-content, and distortion occurs in this structure range with increasing quenching rate. This distortion is more predominant in case of As50Se50 glass. Results of partial pair distribution function show that, most of short range order is mainly governed by the chemically ordered network model, and the random covalent network model plays a relatively small role. Also, it was clarified that the percentage of homopolar bonds increases with increasing quenching rate. In case of x=50at.%, the tendency of As segregation and formation of As–As bonds is demonstrated. Some Deviation from “8-N” rule was observed in the structure of the investigated glasses.

Epidemic spreading dynamics on two-layer complex networks

Recent studies have shown that many real-world systems can be described by multi-layer complex networks. In this paper, the concept of layers is introduced to construct a traffic-driven SIR epidemic spreading model on “logical-physical” layered network. Based on the peak density of infected nodes and the ultimate density of recovered nodes, we investigate the features of epidemic spreading on layered network. Through numerical simulations, it is shown that traffic flow greatly influences the intensity and scope of epidemic spreading. By comparing the effects of four kinds of two-layer networks generated by ER random network model and BA scale-free network model on epidemic spreading, we found that the homogeneity of logical or physical network structure can promote the spread of epidemic more than heterogeneous networks. This work may be of service to design traffic-driven epidemic prevention and control strategies.

Random Network Model for Assessing the Topological Performance of Water Distribution Systems

A complex network approach enables identifying various topological characteristics of critical infrastructure such as water distribution networks (WDNs). These infrastructure networks are designed based on site-specific conditions and also change over time, so at some point, their topological properties may require improvement, especially in terms of their functions. This research aims to analyze WDNs by establishing a random network with varying proportions of grids (pg). To generate random networks, the properties of the real WDN of a city in South Korea were used as a reference point. Our results indicate that, as the WDN shifted from a loop to a branched structure, the system functioning changed non-linearly with the existence of a threshold below which a network may be effectively improved with low resource inputs. More specifically, the threshold pgs were identified as 0.40 and 0.44 for network efficiency and decentralization. Additionally, the threshold pgs were around 0.22 and 0.23 with respect to availability and normalized efficiency when vulnerability was assessed by removing 1% of nodes whereas those were 0.40 and 0.45 when 10% of nodes were removed. On the other hand, the pg of the real WDN was 0.02 which was significantly lower than the thresholds, resulting in low network efficiency, more centralization and increased vulnerability. This suggests that an increase in pg is required to improve the performance of the WDN in the target area. Our study provides a novel framework for identifying a threshold for the functioning of WDNs. This threshold can be used for designing a new network, as well as for improving existing networks for better topological performance.

Understanding networks with exponential-family random network models

The structure of many complex social networks is determined by nodal and dyadic covariates that are endogenous to the tie variables. While exponential-family random graph models (ERGMs) have been very successful in modeling social networks with exogenous covariates, they are often misspecified for networks where some covariates are stochastic. Exponential-family random network models (ERNMs) are an extension of ERGM that retain the desirable properties of ERGM, but allow the joint modeling of tie variables and covariates. We compare ERGM to ERNM to show how conclusions of ERGM modeling are improved by consideration of the ERNM framework. In particular, ERNM simultaneously represents the effects of social influence and social selection processes, while commonly used models do not.

Incorporation of neighborhood information improves performance of SDB models

Mapping of water depth in nearshore waters is important for safe navigation and other human uses of coastal areas. Optical satellite imagery offers a cost-efficient source of data for this purpose. Several methods have been developed for satellite-derived bathymetry (SDB), but the details of their implementation, as well as their relative performance and its dependence on the data and environmental context, is unclear. We compared five SDB models, calibrated and tested against airborne lidar data from southern Corsica. We used a blocked strategy to split data into calibration, validation, and test data, because randomly splitting the data resulted in spatial autocorrelation between these data sets and underestimation of model prediction errors. In general, the non-parametric random forest and neural network models outperformed the parametric multi-band and band-ratio models. The convolutional neural network (CNN) model performed the best, especially in relatively deep (10–20 m) areas, and it could be near-optimally tuned with only ∼1000 data points. However, developing this CNN model was an iterative trial-and-error process that took several months. Using the random forest model, we demonstrated that using information from more than two bands in the visible and near-infrared spectrum contributed to improving model performance, as did incorporation of information from the local neighborhood. We suggest that the use of more than two bands, and the inclusion of spatial contextual information in addition to the values of the individual pixel, should become standard practice in SDB.

Differentiating brain states via multi-clip random fragment strategy-based interactive bidirectional recurrent neural network

EEG is widely adopted to study the brain and brain computer interface (BCI) for its non-invasiveness and low costs. Specifically EEG can be applied to differentiate brain states, which is important for better understanding the working mechanisms of the brain. Recurrent neural network (RNN)-based learning strategy has been widely utilized to differentiate brain states, because its optimization architectures improve the classification performance for differentiating brain states at the group level. However, present classification performance is still far from satisfactory. We have identified two major focal points for improvements: one is about organizing the input EEG signals, and the other is related to the design of the RNN architecture. To optimize the above-mentioned issues and achieve better brain state classification performance, we propose a novel multi-clip random fragment strategy-based interactive bidirectional recurrent neural network (McRFS-IBiRNN) model in this work. This model has two advantages over previous methods. First, the McRFS component is designed to re-organize the input EEG signals to make them more suitable for the RNN architecture. Second, the IBiRNN component is an innovative design to model the RNN layers with interaction connections to enhance the fusion of bidirectional features. By adopting the proposed model, promising brain states classification performances are obtained. For example, 96.97% and 99.34% of individual and group level four-category classification accuracies are successfully obtained on the EEG motor/imagery dataset, respectively. A 99.01% accuracy can be observed for four-category classification tasks with new subjects not seen before, which demonstrates the generalization of our proposed method. Compared with existing methods, our model outperforms them with superior results. Overall, the proposed McRFS-IBiRNN model demonstrates great superiority in differentiating brain states on EEG signals

Microstructural insights into the enigmatic network of random fibers: van Wyk’s notions revisited

Understanding the structure–property relationships of random materials is a long-standing problem of great interest as their applications span one of the most diverse fields of science and engineering. Polycrystals, polymer blends, foams, fluidized beds, cermets, rocks, fibrous materials, and composites are notable examples of these materials. In random fiber networks, the apparent disconnect between size effects, intricate porous characteristics, and the stochastic nature of constituents, leading to quasi-heterogeneous features, emphasizes the need to explore beyond their microstructure. Herein, we revisit the pioneering work of van Wyk on the architectural-based interfiber spacing model for random fiber networks established via the excluded area and excluded volume approaches. In van Wyk’s original work, the discrepancy between the excluded area and excluded volume frameworks has been ascertained and is now addressed by correlating with classical Bertrand’s paradox. Subsequently, the excluded area and excluded volume approaches of the interfiber spacing model have been modified to account for low-aspect-ratio fibers by including the side-side, end-end, and side-end interactions. van Wyk’s modified interfiber spacing model for low-aspect-ratio fibers has been compared with the other analytical model, developed for random fiber networks. A simple relationship between the maximum volume fraction and aspect ratio has also been deduced for cylindrical fibers. To demonstrate our analytical model’s versatility and wide-ranging applicability, the maximum volume fraction and aspect ratio relationship has been benchmarked with a plethora of analytical models, simulations, and experiments obtained from the literature that resulted in a reasonable agreement.

Archaeology of random recursive dags and Cooper-Frieze random networks

AbstractWe study the problem of finding the root vertex in large growing networks. We prove that it is possible to construct confidence sets of size independent of the number of vertices in the network that contain the root vertex with high probability in various models of random networks. The models include uniform random recursive dags and uniform Cooper-Frieze random graphs.

Evolution of the pore structure-transport relationship during catalyst reduction and sintering studied by integrated multi-scale porosimetry and multi-modal imaging

Catalyst pellet fabrication parameters significantly impact final product performance. Tabletted pellets are complex, hierarchical structures that evolve differently over various levels during subsequent processing. Multi-scale porosimetry and multi-modal imaging can, together, encompass all length-scales involved, and, therefore, fully characterise the evolving pellet structure during catalyst reduction and sintering. A random pore-bond network model has highlighted the key pellet structural features determining mass transport, and, thence, was predictive of the impact on mass transfer of controlled modifications to the void space for reduced and aged catalysts. Particular macroporosity, newly induced by reduction and sintering, was critical to mass transport out of proportion to its pore volume fraction. Combined X-ray tomography imaging and percolation modelling showed that reduction and sintering leads to a change (compared to the fresh state) in the initial pellet fabrication parameter that controls mass transport in pellets formed with roll-compacted feed.

Role of composition in fracture behavior of two-phase solids.

In a two-phase solid, we examine the growth of a preexisting macroscopic crack based on simulations of a random spring network model. We find that the enhancement in toughness, as well as strength, is strongly dependent on the ratio of elastic moduli as well as on the relative proportion of the phases. We find that the mechanism that leads to enhancement in toughness is not the same as that for enhancement in strength; however, the overall enhancement is similar in mode I and mixed-mode loading. Based on the crack paths, and the spread of the fracture process zone, we identify the type of fracture to transition from nucleation type, for close to single-phase compositions, whether hard or soft, to avalanche type for more mixed compositions. We also show that the associated avalanche distributions exhibit power-law statistics with different exponents for each phase. The significance of variations in the avalanche exponents with the relative proportion of phases and possible connections to the fracture types are discussed in detail.

Evaluation of the dataset quality in gamma passing rate predictions using machine learning methods.

Gamma passing rate (GPR) predictions using machine learning methods have been explored for treatment verification of radiotherapy plans. However, these methods presented datasets with unbalanced number of plans having different treatment conditions (heterogeneous datasets), such as anatomical sites or dose per fractions, leading to lower model interpretability and prediction performance. We investigated the impact of the dataset composition on GPR binary classification (pass/fail) using random forest (RF), XG-boost, and neural network (NN) models. 945 plans were used to create one reference dataset (randomly assembled) and 24 customized datasets that considered four heterogeneity factors independently (anatomical region, number of arcs, dose per fraction, and treatment unit). 309 predictor features were extracted and calculated from plan parameters, modulation complexity metrics, and radiomic analysis (leave-trajectory maps, 3D dose distributions, and portal dosimetry images). The models' performances were measured using the area under the curve from the receiver operating characteristic (ROC-AUC). Radiomics features for reference models increased ROC-AUC values up to 13%, 15%, and 5% for RF, XG-Boost, and NN, respectively. The datasets with higher heterogeneous conditions presented the lower ROC-AUC values (RF: 0.72 ± 0.11, XG-Boost: 0.67 ± 0.1, NN: 0.89 ± 0.05) compared to models with less heterogeneous treatment conditions (RF: 0.88 ± 0.06, XG-Boost: 0.89 ± 0.07, NN: 0.98 ± 0.01). The ten most important features for each heterogeneity dataset group demonstrated their correlation with the treatments' physical aspects and GPR prediction. Improvements in data generalization and model performances can be associated with datasets having similar treatment conditions. This analysis might be implemented to evaluate the dataset quality and model consistency of further ML applications in radiotherapy. Dataset heterogeneities decrease ML model performance and reliability.

System Modelling and Identification for EEG Monitoring using Random Vector Functional Link Network

Brain signal research occupies a special position in recent biomedical research in recent times. In this work, the authors try to develop a model for monitoring the EEG signal of the patient. It is the extrinsic application of the system identification problem. The Random Vector functional link network (RVFLN) model as the variant of Neural Network, is proposed for the dynamic modeling of a practical system. RVFLN is a fast-learning feed-forward network and does not need iterative tuning that reduces the model's computational complexity and faster training performance. The model is verified with Electroencephalogram (EEG) signal for identification so that it is well suitable for tracking and monitoring systems for patients. The performance of RVFLN is compared with existing models. From the result analysis, it is found that the performance of the proposed RVFLN is most impressive with an efficiency of 99.86%.

Identifying Potential Super-Spreaders and Disease Transmission Hotspots Using White-Tailed Deer Scraping Networks.

White-tailed deer (Odocoileus virginianus, WTD) spread communicable diseases such the zoonotic coronavirus SARS-CoV-2, which is a major public health concern, and chronic wasting disease (CWD), a fatal, highly contagious prion disease occurring in cervids. Currently, it is not well understood how WTD are spreading these diseases. In this paper, we speculate that "super-spreaders" mediate disease transmission via direct social interactions and indirectly via body fluids exchanged at scrape sites. Super-spreaders are infected individuals that infect more contacts than other infectious individuals within a population. In this study, we used network analysis from scrape visitation data to identify potential super-spreaders among multiple communities of a rural WTD herd. We combined local network communities to form a large region-wide social network consisting of 96 male WTD. Analysis of WTD bachelor groups and random network modeling demonstrated that scraping networks depict real social networks, allowing detection of direct and indirect contacts, which could spread diseases. Using this regional network, we model three major types of potential super-spreaders of communicable disease: in-degree, out-degree, and betweenness potential super-spreaders. We found out-degree and betweenness potential super-spreaders to be critical for disease transmission across multiple communities. Analysis of age structure revealed that potential super-spreaders were mostly young males, less than 2.5 years of age. We also used social network analysis to measure the outbreak potential across the landscape using a new technique to locate disease transmission hotspots. To model indirect transmission risk, we developed the first scrape-to-scrape network model demonstrating connectivity of scrape sites. Comparing scrape betweenness scores allowed us to locate high-risk transmission crossroads between communities. We also monitored predator activity, hunting activity, and hunter harvests to better understand how predation influences social networks and potential disease transmission. We found that predator activity significantly influenced the age structure of scraping communities. We assessed disease-management strategies by social-network modeling using hunter harvests or removal of potential super-spreaders, which fragmented WTD social networks reducing the potential spread of disease. Overall, this study demonstrates a model capable of predicting potential super-spreaders of diseases, outlines methods to locate transmission hotspots and community crossroads, and provides new insight for disease management and outbreak prevention strategies.

Model for Medium-Range Order in Metallic Glasses

A new model of the medium-range ordered structure in metallic glasses is proposed. The molecular dynamics simulations revealed that there are two different types of basic clusters formed in metallic glasses. One is the icosahedral cluster (I-cluster) formed around the smaller-sized elements, and the other is the Z14, Z15, and Z16 Frank-Kasper clusters (Z-clusters) formed around the bigger-sized elements. The I- and Z-clusters are mainly connected by volume-sharing type connection and form a network extended to the whole glassy phase. Among them, the connection between Z-clusters corresponds to the dicsclination line proposed by Nelson. The topological nature of the network formed by the Z-clusters can be understood by the continuous random network model.

Classification of Organic and Conventional Vegetables Using Machine Learning: A Case Study of Brinjal, Chili and Tomato

Growing organic food is becoming a challenging task with increasing demand. Food fraud activity has increased considerably with the increase in population growth. Consumers cannot visually distinguish between conventional and organically grown food products. Spectroscopic methodologies are presented to identify chemicals in food, thereby identifying organic and conventional food. Such spectroscopic techniques are laboratory-based, take more time to produce an outcome, and are costlier. Thus, this research designed a portable, low-cost multispectral sensor system to discriminate between organic and conventional vegetables. The designed multispectral sensor system uses a wavelength range (410 nm-940 nm) that includes three bands, namely visible (VIS), ultraviolet (UV) and near-infrared (NIR) spectra, to enhance the accuracy of detection. Tomato, brinjal and green chili samples are employed for the experiment. The organic and conventional discrimination problem is formulated as a classification problem and solved through random forest (RF) and neural network (NN) models, which achieve 92% and 89% accuracy, respectively. A two-stage enhancement mechanism is proposed to improve accuracy. In the first stage, the fuzzy logic mechanism generates additional feature sets. Ant colony optimization (ACO) algorithm-based parameter tuning and feature selection are employed in the second stage to enhance accuracy further. This two-stage improvement mechanism results in 100% accuracy in discriminating between organic and conventional vegetable samples. The detected adulterant is displayed on a web page through an IoT-developed application module to be accessed from anywhere.

A hybrid neural network and optimization algorithm for forecasting and trend detection of Forex market indices

An Autoencoder (AE) is an independent feature extractor from data samples and a deep network can be obtained by stacking several AEs. This paper presents a novel hybrid stacked Autoencoder-based Deep Kernel-based Random Vector Functional Link Network (DKRVFLN-AE) for forecasting and trend analysis of Foreign Exchange (Forex) rates. The proposed model dispenses the random choices of weights and biases, unlike the stacked Deep Random Vector Functional Link Network-AE (DRVFLN-AE), using a wavelet kernel function with strong data fitting capability based on Mercer’s condition. A modified metaheuristic Water Cycle Algorithm is used to optimize the wavelet kernel parameters and provide DKRVFLN-AE a better generalization and learning capability, faster execution speed, lower storage space, and improved accuracy to traditional deep learning Random Vector Functional Link Network and Extreme Learning Machine models. Applications of this new approach to predict exchange rates and trend analysis on three foreign exchange markets provide successful results and validate its superiority over well-known approaches like Random Vector Functional Link Networks, Support Vector Machines, Naive-Bayes, Extreme Learning Machines, and Deep Random Vector Functional Link Network.

Influence of Support Parameter and Excavation Methods on Statistical Distribution Characteristics of Surrounding Rock Pressure in Shallow-Buried Metro Tunnel

For large-spanshallow-buried tunnels in composite strata of urban metro, the study on the distribution characteristics of surrounding rock pressure is a necessary prerequisite for the reliability analysis of tunnel structure. In this paper, through numerical simulation, the random joint file generated by MATLAB is combined with the discrete element software to establish the corresponding random network model. According to the network model calculation, the influence of surrounding rock pressure on the support structure system is discussed, and the reasonable support parameter types are given. Comparing the whole section excavation with the subsectional excavation, the subsectional excavation can effectively control the internal stress self-adjustment of the surrounding rock and effectively restrain the plastic zone. It provides some theoretical guidance and reference for tunnel reliability design.