Network Modeling Framework Research Articles

Crowding and attention in a framework of neural network model.

In this article, I present a framework that would accommodate the classic ideas of visual information processing together with more recent computational approaches. I used the current knowledge about visual crowding, capacity limitations, attention, and saliency to place these phenomena within a standard neural network model. I suggest some revisions to traditional mechanisms of attention and feature integration that are required to fit better into this framework. The results allow us to explain some apparent theoretical controversies in vision research, suggesting a rationale for the limited spatial extent of crowding, a role of saliency in crowding experiments, and several amendments to the feature integration theory. The scheme can be elaborated or modified by future research.

Effect of APOE \u03b54 on multimodal brain connectomic traits: a persistent homology study

BackgroundAlthough genetic risk factors and network-level neuroimaging abnormalities have shown effects on cognitive performance and brain atrophy in Alzheimer’s disease (AD), little is understood about how apolipoprotein E (APOE) ε4 allele, the best-known genetic risk for AD, affect brain connectivity before the onset of symptomatic AD. This study aims to investigate APOE ε4 effects on brain connectivity from the perspective of multimodal connectome.ResultsHere, we propose a novel multimodal brain network modeling framework and a network quantification method based on persistent homology for identifying APOE ε4-related network differences. Specifically, we employ sparse representation to integrate multimodal brain network information derived from both the resting state functional magnetic resonance imaging (rs-fMRI) data and the diffusion-weighted magnetic resonance imaging (dw-MRI) data. Moreover, persistent homology is proposed to avoid the ad hoc selection of a specific regularization parameter and to capture valuable brain connectivity patterns from the topological perspective. The experimental results demonstrate that our method outperforms the competing methods, and reasonably yields connectomic patterns specific to APOE ε4 carriers and non-carriers.ConclusionsWe have proposed a multimodal framework that integrates structural and functional connectivity information for constructing a fused brain network with greater discriminative power. Using persistent homology to extract topological features from the fused brain network, our method can effectively identify APOE ε4-related brain connectomic biomarkers.

Automatic Recognition of Student Emotions Based on Deep Neural Network and Its Application in Depression Detection

The mental health rating scales combined with psychological expert consultation are easily influenced by subjective factors from psychological experts and lack objectivity and scientificalness. Due to the serious influence of depression tendency on learning and living of college students, a novel DNN network model framework based on context emotion information is designed to achieve the automatic auxiliary emotion classification. More corpus training sample can be received by heightening the sample length without reducing samples' number. Firstly, the existing sample feature is input into the recognition model to encode depression related features, and then the MADN features of the samples in the two adjacent segments are input into the above trained model in order for fine-tuning and optimization. Compared with the existing optimal method, the proposed model improves the recognition accuracy in the diagnosis of depression. From the analysis of the experimental results, it is known that deep learning network can monitor the emotional state of college students with high precision, which can accurately identify the patients with depression. The deep learning model can take effective measures to prevent the depression of college students, and discover the depression of college students, alleviate and treat the depression of college students, reduce the depression rate of college students.

Fibrillar Network Dynamics during Oscillatory Rheology of Supramolecular Gels.

Supramolecular gels are three-dimensional network structures formed by the hierarchical self-assembly of small molecules through weak noncovalent interactions. On the basis of the various interactions contributed by specific functional groups/moieties, gels with different architectures can be constructed that are smart to the external stimuli such as pH, type of solvent, stress, temperature, etc. In the present work, we explore the oscillatory shear response of supramolecular self-assembled systems formed by the low-molecular-weight (LMW) gelator based on difunctionalized amino acid, florenylmethoxycarbonyl (Fmoc)-lysine(Fmoc), Fm-K(Fm) in aqueous buffer solution, at two different pH (6 and 7.4). Small amplitude oscillatory shear (SAOS) reported weak frequency dependence of moduli indicating a gel-like network structure. Large amplitude oscillatory shear (LAOS) indicated flow regimes dictated by yielding and subsequent network dynamics analogous to cagelike soft glassy events reported for colloidal systems. The three interval thixotropy test (3iTT) indicated recovery of moduli due to regelation contributed by the reversible interactions. A generalized network model framework is utilized to investigate the transient network characteristics of the Fm-K(Fm) gels. The "network creation" and "network loss" rates were chosen as exponential functions of scaled shear stress (= ) to effectively describe the complex response. On the basis of the insights, possible mechanisms to explain the differences/similarities in the response at different pH are speculated. It is further illustrated that the modeling strategy can be extended to supramolecular gels of different classes because of the commonality/universality of their response features under oscillatory shear.

Hazard Mapping of the Rainfall–Landslides Disaster Chain Based on GeoDetector and Bayesian Network Models in Shuicheng County, China

Landslides are among the most frequent natural hazards in the world. Rainfall is an important triggering factor for landslides and is responsible for topples, slides, and debris flows—three of the most important types of landslides. However, several previous relevant research studies covered general landslides and neglected the rainfall–topples–slides–debris flows disaster chain. Since landslide hazard mapping (LHM) is a critical tool for disaster prevention and mitigation, this study aimed to build a GeoDetector and Bayesian network (BN) model framework for LHM in Shuicheng County, China, to address these geohazards. The GeoDetector model will be used to screen factors, eliminate redundant information, and discuss the interaction between elements, while the BN model will be used for constructing a causality disaster chain network to determine the probability and risk level of the three types of landslides. The practicability of the BN model was confirmed by error rate and scoring rules validation. The prediction accuracy results were tested using overall accuracy, Matthews correlation coefficient, relative operating characteristics curve, and seed cell area index. The proposed framework is demonstrated to be sufficiently accurate to construct the complex LHM. In summary, the combination of the GeoDetector and BN model is very promising for spatial prediction of landslides.

Collagen–iron oxide nanoparticle based ferrogel: large reversible magnetostrains with potential for bioactuation

Smart materials such as stimuli responsive polymeric hydrogels offer unique possibilities for tissue engineering and regenerative medicine. As, however, most synthetic polymer systems and their degradation products lack complete biocompatibility and biodegradability, this study aims to synthesize a highly magnetic responsive hydrogel, based on the abundant natural biopolymer collagen. As the main component of vertebratal extracellular matrix, it reveals excellent biocompatibility. In combination with incorporated magnetic iron oxide nanoparticles, a novel smart nano-bio-ferrogel can be designed. While retaining its basic biophysical properties and interaction with living cells, this collagen-nanoparticle hydrogel can be compressed to 38% of its original size and recovers to 95% in suitable magnetic fields. Besides the phenomenology of this scenario, the underlying physical scenarios are also discussed within the framework of network models. The observed reversible peak strains as large as 150% open up possibilities for the fields of biomedical actuation, soft robotics and beyond.

Prediction of residual saturation and pressure drop during coalescence filtration using dynamic pore network model

Microstructure optimization of mist and dust filter media used in many applications including respirators, lubricated machining, compressors and engine crankcases is increasingly becoming reliant on computational simulations of dry and wet filtration processes. Computational fluid dynamics (CFD) based full-morphology simulations with advanced interface capturing techniques offer unique advantages in enabling accurate characterization of the pore-scale transport processes, yet demand substantial computational effort especially for fine filters. In this work, a dynamic pore network modeling (DPNM) framework is developed and evaluated for the prediction of residual saturation and pressure drop during mist filtration, and applied to an oil-air system. Two nonwoven filters (stainless steel and polyester) with different fibre diameters and packing densities are considered. The pore network is extracted using a watershed-based algorithm, from realistic virtual filter geometries (Abishek et al., 2017), and validated for macroscopic properties such as packing density. The validity of the steady PNM model is assessed by comparison of the dry (air) pressure drop against CFD and experimental data in the literature. The validity of the DPNM model is assessed by comparison of multiphase pressure drops and residual saturation against micro-CT and gravimetric experiments carried out in this work, as well as pore-scale CFD simulations using the volume-of-fluid model. It is found that the dynamic pore network model is able to successfully predict the critical properties of the filtration process such as saturation (local and total) and pressure drop to within 15% of experiment and CFD. Considering that the overall computational cost of DPNM is about 2–3 orders of magnitude lower than fully resolved CFD, DPNM stands as a potential tool for the early-stage design and optimization of mist filtration media.

Optimal charging scheduling and management for a fast-charging battery electric bus system

Battery electric buses (BEBs) are rapidly being embraced by public transit agencies because of their environmental and economic benefits. To address the problems of limited driving range and time-consuming charging for BEBs, manufacturers have developed rapid on-route charging technology that utilizes typical layovers at terminals to charge buses in operation using high power. With on-route fast-charging, BEBs are as capable as their diesel counterparts in terms of range and operating time. However, on-route fast-charging makes it more challenging to schedule and manage charging events for a BEB system. First, on-route fast-charging may lead to high electricity power demand charges. Second, it may increase electricity energy charges because of charging that occurs during on-peak hours. Without careful charging scheduling and management, on-route fast-charging may significantly increase fuel costs and reduce the economic attractiveness of BEBs. The present study proposes a network modeling framework to optimize the charging scheduling and management for a fast-charging BEB system, effectively minimizing total charging costs. The charging schedule determines when to charge a BEB, while the charging management strategically controls the actual charging power. Charging costs include both electricity demand charges and energy charges. The charging scheduling and management problem is first formulated as a nonlinear nonconvex program with time-continuous variables. A discretizing method and a linear reformulation technique are then adopted to reformulate the model as a linear program, which can be easily solved using off-the-shelf solvers, even for large-scale problems. Finally, the model is demonstrated with extensive numerical studies based on two real-world bus networks. The results demonstrate that the proposed model can effectively determine the optimal charging scheduling and management for a fast-charging BEB system, which carries the potential for use in large-scale real-world bus networks.

A Guard Cell Abscisic Acid (ABA) Network Model That Captures the Stomatal Resting State.

Stomatal pores play a central role in the control of carbon assimilation and plant water status. The guard cell pair that borders each pore integrates information from environmental and endogenous signals and accordingly swells or deflates, thereby increasing or decreasing the stomatal aperture. Prior research shows that there is a complex cellular network underlying this process. We have previously constructed a signal transduction network and a Boolean dynamic model describing stomatal closure in response to signals including the plant hormone abscisic acid (ABA), calcium or reactive oxygen species (ROS). Here, we improve the Boolean network model such that it captures the biologically expected response of the guard cell in the absence or following the removal of a closure-inducing signal such as ABA or external Ca2+. The expectation from the biological system is reversibility, i.e., the stomata should reopen after the closing signal is removed. We find that the model’s reversibility is obstructed by the previously assumed persistent activity of four nodes. By introducing time-dependent Boolean functions for these nodes, the model recapitulates stomatal reopening following the removal of a signal. The previous version of the model predicts ∼20% closure in the absence of any signal due to uncertainty regarding the initial conditions of multiple network nodes. We systematically test and adjust these initial conditions to find the minimally restrictive combinations that appropriately result in open stomata in the absence of a closure signal. We support these results by an analysis of the successive stabilization of feedback motifs in the network, illuminating the system’s dynamic progression toward the open or closed stomata state. This analysis particularly highlights the role of cytosolic calcium oscillations in causing and maintaining stomatal closure. Overall, we illustrate the strength of the Boolean network modeling framework to efficiently capture cellular phenotypes as emergent outcomes of intracellular biological processes.

Application of mechanistic modelling and machine learning for cream cheese fermentation pH prediction

AbstractBACKGROUNDCheese production occupies a small but growing share of the current dairy industry. Industrial cream cheese production involves the complex scheduling of multiple batch fermenters and various downstream units. However, significant end‐time variation from different fermentation batches makes downstream scheduling challenging and thus decreases the process throughput.RESULTSThis research addressed this challenge by using an artificial neural network (a Long‐Short Term Memory Network, LSTM) in combination with a mechanistic model describing the changes in biomass, lactose, and lactic acid concentrations. The LSTM network/mechanistic modelling approach shows an end‐time difference of 3 min over batch times of 6 to 7 h compared to the laboratory experiment, with an overall accuracy (R2) of over 0.99.CONCLUSIONThe proposed hybrid fundamental/artificial neural network (ANN) model framework could reasonably predict the cheese fermentation pH with limited data. The outcome of this research enables fermentation end‐time prediction, makes the downstream scheduling possible, and thus, could help to improve the process throughput. The developed model could also be used for further cheese digital twin development, based on how better product quality control and higher process efficiency could be achieved. © 2020 Society of Chemical Industry

Artificial neural network modeling of sliding wear

A widely used type of artificial neural networks, called multilayer perceptron, is applied for data-driven modeling of the wear coefficient in sliding wear under constant testing conditions. The integral and differential forms of wear equation are utilized for designing an artificial neural network-based model for the wear rate. The developed artificial neural network modeling framework can be utilized in studies of wearing-in period and the so-called true wear coefficient. Examples of the use of the developed approach are given based on the experimental data published recently.

Personalized Network Modeling of the Pan-Cancer Patient and Cell Line Interactome.

PURPOSEPersonalized network inference on diverse clinical and in vitro model systems across cancer types can be used to delineate specific regulatory mechanisms, uncover drug targets and pathways, and develop individualized predictive models in cancer.METHODSWe developed TransPRECISE (personalized cancer-specific integrated network estimation model), a multiscale Bayesian network modeling framework, to analyze the pan-cancer patient and cell line interactome to identify differential and conserved intrapathway activities, to globally assess cell lines as representative models for patients, and to develop drug sensitivity prediction models. We assessed pan-cancer pathway activities for a large cohort of patient samples (> 7,700) from the Cancer Proteome Atlas across ≥ 30 tumor types, a set of 640 cancer cell lines from the MD Anderson Cell Lines Project spanning 16 lineages, and ≥ 250 cell lines’ response to > 400 drugs.RESULTSTransPRECISE captured differential and conserved proteomic network topologies and pathway circuitry between multiple patient and cell line lineages: ovarian and kidney cancers shared high levels of connectivity in the hormone receptor and receptor tyrosine kinase pathways, respectively, between the two model systems. Our tumor stratification approach found distinct clinical subtypes of the patients represented by different sets of cell lines: patients with head and neck tumors were classified into two different subtypes that are represented by head and neck and esophagus cell lines and had different prognostic patterns (456 v 654 days of median overall survival; P = .02). High predictive accuracy was observed for drug sensitivities in cell lines across multiple drugs (median area under the receiver operating characteristic curve > 0.8) using Bayesian additive regression tree models with TransPRECISE pathway scores.CONCLUSIONOur study provides a generalizable analytic framework to assess the translational potential of preclinical model systems and to guide pathway-based personalized medical decision making, integrating genomic and molecular data across model systems.

On the Importance of Incorporating Structural Heterogeneity of Porous Electrodes in LI-ION Battery Models: Pore Network Modelling a Way to MOVE Forward

Image processing of 3D tomographic images to extract structural information of porous materials has become extremely important in porous media research with the commoditization of x-ray tomography equipment to the lab scale. Extracted pore networks from image analysis techniques enable transport properties calculation for bigger domains at a very low computational cost, allowing substantial investigation of porous media. Consequently, they are being increasingly used to model Multiphysics transport processes in various energy storage devices like fuel cells and batteries, where the geometrical structure plays a vital role in electrochemical performance, but true pore-scale modeling is computationally infeasible.Traditionally, pore network modelling has been used to find electrochemical performance of Li Ion batteries by investigating pore phase interconnectivity in carbon electrodes. The actual lithium ion electrode, however, consists of active material, carbon binder and electrolyte filled pore phase. The interconnectivity of these phases influences the effective transport properties and hence electrochemical performance of cathode material. The presence of carbon binder phase not only reduces the SEI layer between active material and electrolyte phase but also influences tortuosity and effective electronic conductivity in the pore and solid phase respectively. The present work uses a pore network modelling frame work to investigate the effect of carbon binder phase on Lithium Ion battery cathode. We used actual three phase, X-ray tomography image of NMC-811 cathode material and studied its electrochemical performance with and without binder phase. Unlike previous models which compensate the importance of carbon binder as electrical conductor by assuming high electronic conductivity of active material, this study considers actual values of conductivities in all phases. Moreover, the impact of nanoporosity in the carbon binder phase was also explored and found to enhance the reaction rate compared to solid binder. The reduction in computational time achieved using the pore-network approach was so significant that adding additional physics and transient conditions could conceivably be included to increase the accuracy of the model, thereby providing more realistic simulations and point to the true limiting processes. The developed pore network model opens a new avenue for modelling complex electrochemical systems with less computational cost, enabling simulation of bigger electrode domains while keeping structural heterogeneities of material.

Ensemble Deep Learning Models for Heart Disease Classification: A Case Study from Mexico

Heart diseases are highly ranked among the leading causes of mortality in the world. They have various types including vascular, ischemic, and hypertensive heart disease. A large number of medical features are reported for patients in the Electronic Health Records (EHR) that allow physicians to diagnose and monitor heart disease. We collected a dataset from Medica Norte Hospital in Mexico that includes 800 records and 141 indicators such as age, weight, glucose, blood pressure rate, and clinical symptoms. Distribution of the collected records is very unbalanced on the different types of heart disease, where 17% of records have hypertensive heart disease, 16% of records have ischemic heart disease, 7% of records have mixed heart disease, and 8% of records have valvular heart disease. Herein, we propose an ensemble-learning framework of different neural network models, and a method of aggregating random under-sampling. To improve the performance of the classification algorithms, we implement a data preprocessing step with features selection. Experiments were conducted with unidirectional and bidirectional neural network models and results showed that an ensemble classifier with a BiLSTM or BiGRU model with a CNN model had the best classification performance with accuracy and F1-score between 91% and 96% for the different types of heart disease. These results are competitive and promising for heart disease dataset. We showed that ensemble-learning framework based on deep models could overcome the problem of classifying an unbalanced heart disease dataset. Our proposed framework can lead to highly accurate models that are adapted for clinical real data and diagnosis use.

Network modeling and Internet of things for smart and connected health systems—a case study for smart heart health monitoring and management

Heart disease is a leading cause of death in the US. Recent advances in the Internet of Things (IoT) provide a great opportunity to realize smart and connected health systems through IoT monitoring and sensor-based data analytics of cardiac disorders. However, big data arising from the large-scale IoT system pose a significant challenge for efficient and effective sensory information processing and decision making. Very little has been done to glean pertinent information about the disease-altered cardiac activity in the context of large-scale IoT network. In this study, we propose a parallel computing framework for multi-level network modeling and monitoring of cardiac dynamics to realize the potential of IoT-enabled smart health management. Specifically, dissimilarities among cardiac signals are firstly characterized among heartbeats for an individual patient, as well as among representative heartbeats for different patients. Then, a stochastic learning approach is developed to optimize the embedding of cardiac signals into a beat-to-beat network model, as well as a patient-to-patient network model. Further, we develop a parallel computing algorithm to improve the computational efficiency. Finally, a statistical process monitoring scheme is designed to harness network features for real-time monitoring and anomaly detection of cardiac activities. Experimental results show the proposed methodology has strong potential to realize a smart and interconnected system for cardiac health management in the context of large-scale IoT network.

Hierarchical Bayesian Network Modeling Framework for Large-Scale Process Monitoring and Decision Making

In this brief, a hierarchical Bayesian network modeling framework is formulated for large-scale process monitoring and decision making, which includes a basic layer and a functional layer. First, the whole process is decomposed into different units, where local Bayesian networks are constructed, providing monitoring information and decision-making capability for the upper layer. The network structure is determined automatically based on the process data in each local unit of the basic layer. Then, through incorporating the topological structure of the process, a functional Bayesian network is further constructed to infer the information from the basic layer, which can be customized according to user demands, such as fault detection, fault diagnosis, and classification of operating status. The performance of the proposed method is evaluated through a benchmark process.

Novel Regional Activity Representation With Constrained Canonical Correlation Analysis for Brain Connectivity Network Estimation.

Inferring brain connectivity networks from fMRI data can take place at the Region of Interest (ROI) or voxel level. With most ROI-based approaches, the signals from same-ROI voxels are simply averaged, neglecting any inhomogeneity in each ROI and assuming that the same voxels will interact with different ROIs in a similar manner. In this paper, we propose a novel method of representing ROI activity and estimating brain connectivity that takes into account the regionally-specific nature of brain activity, the spatial location of concentrated activity, and activity in other ROIs. The proposed method is able to integrate intrinsic regional structures into a network modelling framework, which we call local activity constrained canonical correlation analysis (LA-cCCA). We evaluated LA-cCCA on both simulated and real fMRI data. The simulation results demonstrated that LA-cCCA had improved accuracy of the estimated brain connectivity networks compared to the average-signal or Principal Component Analysis (PCA)-based correlation methods and the Canonical Correlation Analysis (CCA) method. We further examined the performance of LA-cCCA on real fMRI data set from the Human Connectome Project. LA-cCCA outperformed the other three approaches in terms of connectivity reproducibility. The proposed method explores the potentials of regional activity representation and is a reliable model for connectivity network estimation. It may serve as a promising tool for studying both the healthy and diseased brain.

Power Delivery Network (PDN) Modeling for Backside-PDN Configurations With Buried Power Rails and $\mu$ TSVs

In this article, a power delivery network (PDN) modeling framework for backside-PDN configurations is presented. A backside-PDN configuration contains dense microthrough silicon vias ( $\mu $ TSVs) and power/ground metal stack on the backside of the die. This approach separates the PDN from a conventional signaling network of the back-end-of-the-line (BEOL) and improves power integrity and core utilization. We benchmark this technology with conventional front-side BEOL PDN configurations. Owing to the lower resistivity compared with Cu metal lines for advanced technology nodes, we use ruthenium (Ru)-based buried power rail for PDN modeling. Our analysis shows that the steady-state IR-drop reduces by more than $4\times $ in the backside-PDN configuration, and a simultaneous switching noise analysis shows a significant reduction in transient droops. The framework results are validated with a place-and-route (P&R)-based physical implementation flow. We quantify the area improvement in the actual flow and observe 25%–30% improvement in the backside-PDN configuration. From a PDN modeling framework, the PDN results follow a trend similar to the ones obtained from the block-level P&R of the given configurations. Moreover, we investigate the impacts of package-to-die interconnect pitch, metal–insulator–metal cap density, and input pulse on the PDN performance. In addition, we perform thermal modeling to analyze the thermal implications of the backside-PDN configuration. From a thermal modeling perspective, there is negligible influence from a dielectric bonding layer in the backside-PDN configuration.

Bayesian modeling of flood control networks for failure cascade characterization and vulnerability assessment

AbstractThis paper presents a Bayesian network model to assess the vulnerability of the flood control infrastructure and to simulate failure cascade based on the topological structure of flood control networks along with hydrological information gathered from sensors. Two measures are proposed to characterize the flood control network vulnerability and failure cascade: (a) node failure probability (NFP), which determines the failure likelihood of each network component under each scenario of rainfall event, and (b) failure cascade susceptibility, which captures the susceptibility of a network component to failure due to failure of other links. The proposed model was tested in both single watershed and multiple watershed scenarios in Harris County, Texas using historical data from three different flooding events, including Hurricane Harvey in 2017. The proposed model was able to identify the most vulnerable flood control network segments prone to flooding in the face of extreme rainfall. The framework and results furnish a new tool and insights to help decision‐makers to prioritize infrastructure enhancement investments and actions. The proposed Bayesian network modeling framework also enables simulation of failure cascades in flood control infrastructures, and thus could be used for scenario planning as well as near‐real‐time inundation forecasting to inform emergency response planning and operation, and hence improve the flood resilience of urban areas.

An assignment model for public transport networks with both schedule- and frequency-based services

This paper presents an assignment modeling framework for public transport networks with co-existing schedule- and frequency-based services. The paper develops, applies and discusses a joint model, which aims at representing the behavior of passengers as realistically as possible. The model consists of a choice set generation phase followed by a multinomial logit route choice model and assignment of flow to the generated alternatives. The choice set generation uses an event dominance principle to exclude alternatives with costs above a certain cost threshold. Furthermore, a heuristic for aggregating overlapping lines is proposed. The results from applying the model to a case study in the Greater Copenhagen Area show that the level of service obtained in the unified network model of mixed services is placed between the level of service for strictly schedule-based and strictly frequency-based networks. The results also show that providing timetable information to the passengers improve their utility function as compared to only providing information on frequencies.