Multi Network Model Research Articles

An Internet of Things Task Scheduling Framework Based on Agile Virtual Network on Demand Service Model

In order to improve the efficiency of cloud computing resource utilization and avoid the problem of computing resource allocation and scheduling lagging behind load changes, the author proposes a cloud computing resource on-demand allocation and elastic scheduling method based on network load prediction. Firstly, the author takes the network load data of Wikimedia as the research object and proposes an adaptive two-stage multi network model load prediction method based on LSTM, i.e. ATSMNN-LSTM load prediction method). This method can classify the network load data into climbing and descending types based on the trend and characteristics of the input network load data, And adaptively schedule the input network load data to the LSTM load prediction model that matches its type for prediction based on the classification results. The author proposes a maximum cloud service revenue computing resource quantity search algorithm based on network load prediction (i.e. MaxCSPRNWP algorithm), which aims to improve cloud service revenue as the optimization objective. Under the premise of ensuring task service quality and system stability, the algorithm allocates cloud computing resources on demand and flexibly schedules them in advance based on the predicted network load results. The experimental results show that the ATSMNNLSTM load prediction method proposed by the author can obtain more accurate network load prediction results compared to other load prediction methods, and the MaxCSPR-NWP algorithm, which is based on network load prediction and is capable of effectively converting the network load prediction results into the required number of cloud servers, is the maximum cloud service revenue computing resource quantity search algorithm proposed by the author, not only does it achieve the early allocation and scheduling of cloud computing resources, thereby avoiding the impact of lagging behind in computing resource allocation and scheduling due to load changes on the quality of cloud computing task services and resource utilization efficiency, at the same time, it has also achieved on-demand allocation and flexible scheduling of cloud computing resources with the goal of improving cloud service revenue.

A multi-network model of Parkinson’s disease tremor: exploring the finger-dimmer-switch theory and role of dopamine in thalamic self-inhibition

Background. Tremor is a cardinal symptom of Parkinson's disease (PD) that manifests itself through complex oscillatory activity across multiple neuronal populations. According to the finger-dimmer-switch (FDS) theory, tremor is triggered by transient pathological activity in the basal ganglia-thalamo-cortical (BTC) network (the finger) and transitions into an oscillatory form within the inner circuitry of the thalamus (the switch). The cerebello-thalamo-cortical (CTC) network (the dimmer) is then involved in sustaining and amplifying tremor amplitude. In this study, we aimed to investigate the generation and progression dynamics of PD tremor oscillations by developing a comprehensive and interacting FDS model that transitions sequentially from healthy to PD to tremor and then to tremor-off state.Methods.We constructed a computational model consisting of 700 neurons in 11 regions of BTC, CTC, and thalamic networks. Transition from healthy to PD state was simulated through modulating dopaminergic synaptic connections; and further from PD to tremor and tremor-off by modulating projections between the thalamic reticular nucleus (TRN), anterior ventrolateral nucleus (VLa), and posterior ventrolateral nucleus (VLp).Results.Sustained oscillations in the frequency range of PD tremor emerged in thalamic VLp (5 Hz) and cerebellar dentate nucleus (3 Hz). Increasing self-inhibition in the thalamus through dopaminergic modulation significantly decreased tremor amplitude.Conclusion/Significance.Our results confirm the mechanistic power of the FDS theory in describing the PD tremor phenomenon and emphasize the role of dopaminergic modulation on thalamic self-inhibition. These insights pave the way for novel therapeutic strategies aimed at reducing the tremor by strengthening thalamic self-inhibition, particularly in dopamine-resistant patients.

Deep neural network homogenization of multiphysics behavior for periodic piezoelectric composites

We present a physically informed deep neural network homogenization theory for identifying homogenized moduli and local electromechanical fields of periodic piezoelectric composites. The theory employs a multi-network model to represent solutions to stress equilibrium and electrostatic partial differential equations in the inclusion and the matrix phases, leading to a more accurate depiction of field variables. Satisfaction of periodicity conditions for hexagonal and cubic symmetries are efficiently tackled using a series of sinusoidal functions with known periods and adjustable parameters. Additionally, the theory applies fully trainable weights to the collocation points. These trainable weights, trained concurrently with the neural network weights, compel the neural networks to enhance their performance when faced with large local stress/deformation gradients. The predictive capability of the proposed theory is illustrated by comparison with finite-element solutions for composites reinforced with unidirectional fiber, spherical particle, or weakened by an ellipsoidal cavity over a wide range of volume fractions.

Redefining Neural Architecture Search of Heterogeneous Multinetwork Models by Characterizing Variation Operators and Model Components.

With neural architecture search (NAS) methods gaining ground on manually designed deep neural networks-even more rapidly as model sophistication escalates-the research trend is shifting toward arranging different and often increasingly complex NAS spaces. In this conjuncture, delineating algorithms which can efficiently explore these search spaces can result in a significant improvement over currently used methods, which, in general, randomly select the structural variation operator, hoping for a performance gain. In this article, we investigate the effect of different variation operators in a complex domain, that of multinetwork heterogeneous neural models. These models have an extensive and complex search space of structures as they require multiple subnetworks within the general model in order to answer different output types. From that investigation, we extract a set of general guidelines whose application is not limited to that particular type of model and are useful to determine the direction in which an architecture optimization method could find the largest improvement. To deduce the set of guidelines, we characterize both the variation operators, according to their effect on the complexity and performance of the model; and the models, relying on diverse metrics which estimate the quality of the different parts composing it.

The Networking Brain: How Extracellular Matrix, Cellular Networks, and Vasculature Shape the In Vivo Mechanical Properties of the Brain.

Mechanically, the brain is characterized by both solid and fluid properties. The resulting unique material behavior fosters proliferation, differentiation, and repair of cellular and vascular networks, and optimally protects them from damaging shear forces. Magnetic resonance elastography (MRE) is a noninvasive imaging technique that maps the mechanical properties of the brain in vivo. MRE studies have shown that abnormal processes such as neuronal degeneration, demyelination, inflammation, and vascular leakage lead to tissue softening. In contrast, neuronal proliferation, cellular network formation, and higher vascular pressure result in brain stiffening. In addition, brain viscosity has been reported to change with normal blood perfusion variability and brain maturation as well as disease conditions such as tumor invasion. In this article, the contributions of the neuronal, glial, extracellular, and vascular networks are discussed to the coarse-grained parameters determined by MRE. This reductionist multi-network model of brain mechanics helps to explain many MRE observations in terms of microanatomical changes and suggests that cerebral viscoelasticity is a suitable imaging marker for brain disease.

Deep learning architectures for diagnosing the severity of apple frog-eye leaf spot disease in complex backgrounds.

In precision agriculture, accurately diagnosing apple frog-eye leaf spot disease is critical for effective disease management. Traditional methods, predominantly relying on labor-intensive and subjective visual evaluations, are often inefficient and unreliable. To tackle these challenges in complex orchard environments, we develop a specialized deep learning architecture. This architecture consists of a two-stage multi-network model. The first stage features an enhanced Pyramid Scene Parsing Network (L-DPNet) with deformable convolutions for improved apple leaf segmentation. The second stage utilizes an improved U-Net (D-UNet), optimized with bilinear upsampling and batch normalization, for precise disease spot segmentation. Our model sets new benchmarks in performance, achieving a mean Intersection over Union (mIoU) of 91.27% for segmentation of both apple leaves and disease spots, and a mean Pixel Accuracy (mPA) of 94.32%. It also excels in classifying disease severity across five levels, achieving an overall precision of 94.81%. This approach represents a significant advancement in automated disease quantification, enhancing disease management in precision agriculture through data-driven decision-making.

Cell Performance Prediction of the Particle Structure of All-Solid-State Batteries By Numerical Simulation and Machine Learning

All-Solid-State Battery (ASSB) is expected to be commercialized for electric vehicle applications. However, the correlation between the electrode structure and the cell performance is unknown, and no design theory has been established. Especially the effect of contact stress on solid-solid interface is important. And the ununiform reaction area cause the exsess current concentration. In our previous study, we developed various simulation model to understand the effect of heterogeneous structure and the effect of volume expansion on dynamic structure change during charging-discharging cycle from the viewpoint of durability. In this study, we developed new simulation model with stress calculation and electrochemical reaction, Moreover, the simulation results were used to create training dataset for machine learning. By using these approaches, we tried to develop new method to predict cell performance from composite electrode structure to cell performance.The stress estimation was performed using the discrete element method (DEM) [1]. The various material parameters are used such as, particle diameter (average and standard deviation), Young's modulus, Poisson's ratio and density. The distribution of the stresses on each particle was calculated from the DEM calculation results and confirmed to be approximately Gaussian.To simulate electrochemical reaction and mass transport in electrode layer, we used multi-network model (MNM)[6]. The schematic image of this model is shown in Fig.1, we used various network grid of active material network, solid electrolyte network and interface network. Basic equations of Li concentration, Li potential and electron potential were calculated in these networks. In this study, with DEM and MNM, we developed new simulation model shown in Fig.2.A convolutional neural network (CNN) [7] was used as the model for machine learning, and various structural parameters, including the DOD, cell potential, and stress on each particle calculated by DEM, were identified to create a dataset. 250 images were used for training, 150 of which were used for training, 50 for validation, and 50 for testing. The coefficient of determination (R²) of the standard deviation of DOD and potential is almost 0.7 in each condition, and the cell performance can be predicted by the machine learning model. The contact area of the electrochemical reaction interface, AM-SE, was calculated from the stress on the contact surface calculated by DEM and Hertzian contact theory [8]. Thus, we can express the relationship between interface mechanical stress on solid-solid and effective reaction, and this effect can be predicted with various structure of composite electrode layer by using machine learning. It was possible to predict the cell performance and effective contact area in the electrode layer of the all-solid-state battery.Acknowledgement:This research was carried out under the project (Grant Number JPMJMI21G4) supported by the JST-Mirai Program, Japan.

Network-based identification and mechanism exploration of active ingredients against Alzheimer’s disease via targeting endoplasmic reticulum stress from traditional chinese medicine

Alzheimer's disease is a neurodegenerative disease that leads to dementia and poses a serious threat to the health of the elderly. Traditional Chinese medicine (TCM) presents as a promising novel therapeutic therapy for preventing and treating dementia. Studies have shown that natural products derived from kidney-tonifying herbs can effectively inhibit AD. Furthermore, endoplasmic reticulum (ER) stress is a critical factor in the pathology of AD. Regulation of ER stress is a crucial approach to prevent and treat AD. Thus, in this study, we first collected kidney-tonifying herbs, integrated chemical ingredients from multiple TCM databases, and constructed a comprehensive drug-target network. Subsequently, we employed the endophenotype network (network proximity) method to identify potential active ingredients in kidney-tonifying herbs that prevented AD via regulating ER stress. By combining the predicted outcomes, we discovered that 32 natural products could ameliorate AD pathology via regulating ER stress. After a comprehensive evaluation of the multi-network model and systematic pharmacological analyses, we further selected several promising compounds for in vitro testing in the APP-SH-SY5Y cell model. Experimental results showed that echinacoside and danthron were able to effectively reduce ER stress-mediated neuronal apoptosis by inhibiting the expression levels of BIP, p-PERK, ATF6, and CHOP in APP-SH-SY5Y cells. Overall, this study utilized the endophenotype network to preliminarily decipher the effective material basis and potential molecular mechanism of kidney-tonifying Chinese medicine for prevention and treatment of AD.

Entity recognition method for airborne products metrological traceability knowledge graph construction

The airborne system, as one of the complex and extensive subsystems of an aircraft, primarily performs critical flight assurance functions. The quality of its components has a direct impact on the aircraft's safety and reliability. Metrology documents comprehensively document the performance parameters throughout the entire product life cycle. The Metrological Traceability Knowledge Graph (MTKG) for airborne products offers decision support to engineers engaged in metrological tasks, ensuring the continuous high quality of the products. This paper introduces an entity recognition method for airborne product metrological traceability knowledge graph construction. First, the ontology for MTKG is developed. Next, a fine-tuned multi-network model is proposed. Named entities in the field of metrology are recognized through three stages: word vector representation, sentence feature extraction, and optimal label assignment. Meanwhile, active learning methods are incorporated to reduce the expense of data annotation. The proposed model is validated using an actual metrology corpus, and the experimental results demonstrate its superior performance compared to the other four baseline methods. Finally, the MTKG is developed using this approach, offering engineers intelligent applications, including metrological traceability analysis and traceability path reasoning within the process of product metrology. This enhances the metrology capabilities of airborne products and demonstrates the extensive potential of knowledge graphs in metrology.

Autistic Symptoms, Irritability, and Executive Dysfunctions: Symptom Dynamics from Multi-Network Models.

Socio-cognitive difficulties in individuals with autism spectrum disorder (ASD) are heterogenuous and often co-occur with irritability symptoms, such as angry/grouchy mood and temper outbursts. However, the specific relations between individual symptoms are not well-represented in conventional methods analyzing aggregated autistic symptoms and ASD diagnosis. Moreover, the cognitive-behavioral mechanisms linking ASD to irritability are largely unknown. This study investigated the dynamics between autistic (Social Responsiveness Scale) and irritability (Affective Reactivity Index) symptoms and executive functions (Cambridge Neuropsychological Test Automated Battery) in a sample of children and adolescents with ASD, their unaffected siblings, and neurotypical peers (N = 345, aged 6-18years, 78.6% male). Three complementary networks across the entire sample were computed: (1) Gaussian graphical network estimating the conditional correlations between symptom nodes; (2) Relative importance network computing relative influence between symptoms; (3) Bayesian directed acyclic graph estimating predictive directionality between symptoms. Networks revealed numerous partial correlations within autistic (rs = .07-.56) and irritability (rs = .01-.45) symptoms and executive functions (rs = -.83 to .67) but weak connections between clusters. This segregated pattern converged in all directed and supplementary networks. Plausible predictive paths were found between social communication difficulties to autism mannerisms and between "angry frequently" to "lose temper easily." Autistic and irritability symptoms are two relatively independent families of symptoms. It is unlikely that executive dysfunctions explain elevated irritability in ASD. Findings underscore the need for researching other mood and cognitive-behavioral bridge symptoms, which may inform individualized treatments for co-occurring irritability in ASD.

Application of Physics-Informed Neural Networks for forward and inverse analysis of pile–soil interaction

The application of the Physics-Informed Neural Networks (PINNs) to forward and inverse analysis of pile–soil interaction problems is presented. The main challenge encountered in the Artificial Neural Network (ANN) modeling of pile–soil interaction is the presence of abrupt changes in material properties, which results in large discontinuities in the gradient of the displacement solution. Therefore, a domain-decomposition multi-network model is proposed to deal with the discontinuities in the strain fields at common boundaries of pile–soil regions and soil layers. The application of the model to the analysis and parametric study of single piles embedded in both homogeneous and layered formations is demonstrated under axisymmetric and plane strain conditions. The performance of the model in parameter identification (inverse analysis) of pile–soil interaction is particularly investigated. It is shown that by using PINNs, the localized data acquired along the pile length –possibly obtained via fiber optic strain sensing–can be successfully used for the inversion of soil parameters in layered formations.

Multi-Scale Channel Adaptive Time-Delay Neural Network and Balanced Fine-Tuning for Arabic Dialect Identification

The time-delay neural network (TDNN) can consider multiple frames of information simultaneously, making it particularly suitable for dialect identification. However, previous TDNN architectures have focused on only one aspect of either the temporal or channel information, lacking a unified optimization for both domains. We believe that extracting appropriate contextual information and enhancing channels are critical for dialect identification. Therefore, in this paper, we propose a novel approach that uses the ECAPA-TDNN from the speaker recognition domain as the backbone network and introduce a new multi-scale channel adaptive module (MSCA-Res2Block) to construct a multi-scale channel adaptive time-delay neural network (MSCA-TDNN). The MSCA-Res2Block is capable of extracting multi-scale features, thus further enlarging the receptive field of convolutional operations. We evaluated our proposed method on the ADI17 Arabic dialect dataset and employed a balanced fine-tuning strategy to address the issue of imbalanced dialect datasets, as well as Z-Score normalization to eliminate score distribution differences among different dialects. After experimental validation, our system achieved an average cost performance (Cavg) of 4.19% and a 94.28% accuracy rate. Compared to ECAPA-TDNN, our model showed a 22% relative improvement in Cavg. Furthermore, our model outperformed the state-of-the-art single-network model reported in the ADI17 competition. In comparison to the best-performing multi-network model hybrid system in the competition, our Cavg also exhibited an advantage.

Local Functional Connectivity as a Parsimonious Explanation of the Main Frameworks for ADHD in Medication-Naïve Adults

Objective: Neuroimaging studies in children with ADHD indicate that their brain exhibits an atypical functional connectivity pattern characterized by increased local connectivity and decreased distant connectivity. We aim to evaluate if the local and distant distribution of functional connectivity is also altered in adult samples with ADHD who have never received medication before. Methods: We compared local and distant functional connectivity between 31 medication-naïve adults with ADHD and 31 healthy controls and tested whether this pattern was associated with symptoms severity scores. Results: ADHD sample showed increased local connectivity in the dACC and the SFG and decreased local connectivity in the PCC. Conclusion: Results parallel those obtained in children samples suggesting a deficient integration within the DMN and segregation between DMN, FPN, and VAN. These results are consistent with the three main frameworks that explain ADHD: the neurodevelopmental delay hypothesis, the DMN interference hypothesis and multi-network models.

Local Functional Connectivity as a Parsimonious Explanation of the Main Frameworks for ADHD in Medication-Naïve Adults

Objective: Neuroimaging studies in children with ADHD indicate that their brain exhibits an atypical functional connectivity pattern characterized by increased local connectivity and decreased distant connectivity. We aim to evaluate if the local and distant distribution of functional connectivity is also altered in adult samples with ADHD who have never received medication before. Methods: We compared local and distant functional connectivity between 31 medication-naïve adults with ADHD and 31 healthy controls and tested whether this pattern was associated with symptoms severity scores. Results: ADHD sample showed increased local connectivity in the dACC and the SFG and decreased local connectivity in the PCC. Conclusion: Results parallel those obtained in children samples suggesting a deficient integration within the DMN and segregation between DMN, FPN, and VAN. These results are consistent with the three main frameworks that explain ADHD: the neurodevelopmental delay hypothesis, the DMN interference hypothesis, and multi-network models.

Facial expression recognition in facial occlusion scenarios: A path selection multi-network

The occlusion scenarios including wearing mask, wearing sunglasses, wearing hat, etc. Thus, parts of face are occluded, e.g. nose and mouth are obscured if a mask is worn. Facial expression recognition (FER) tasks have been widely researched. However, less attention has been paid to FER in occlusion scenarios, which are not uncommon in the real world. In this paper, we propose a method that structures a path selection multi-network model to achieve the FER in the above three types of facial occlusion scenarios. The method contains two parts. For the multi-network, we segment the labels in one database which results in three new sub-databases to train three Subnets, respectively. For the path selection, which is an integration method of multi-network, we merge groups of labels in one database to train an initial network called BeginNet. The prediction of BeginNet selects one of the Subnets to make the final prediction. We concatenate four popular databases Fer2013, JAFFE, KDEF and Raf-DB databases into one larger database, and use the combined database to verify our method effectiveness. The experimental results show that our method has better results in coping with the expression recognition task in multiple types of facial occlusion scenarios.

Adaptive nonlinear ANC system based on time-domain signal reconstruction technology

When the primary reference signal obtained by the existing Active noise control (ANC) system is not accurate, the control effect of interior noise will be reduced or even fail. Considering the fault-tolerate and robustness of the system, the study proposes an adaptive nonlinear ANC system for interior noise, which contains noise signal decomposition, multi-network reconstruction model and Variable step-size LMS (VSS-LMS) algorithm. The noise signal decomposition method is used to address the non-stationary of the interior noise; Based on the signal components, the multi-network model for the noise signal reconstruction of passenger ear-sides is designed, which is pre-trained by a restricted Boltzmann machine for improved reconstruction accuracy and realize the adaptive extraction of signal features; And then based on the reconstruction signal components, the controller weights of corresponding components are adaptively updated by the VSS-LMS algorithm to control the passenger ear-sides noise. The effectiveness of the proposed adaptive nonlinear ANC system is validated using noise signal sources collected from a vehicle. Compared with the different ANC systems, the proposed system is superior in terms of fault-tolerant and robustness, which can guarantee stable work of the interior noise control.

Recognition Method of Abnormal Behavior of Marine Fish Swarm Based on In-Depth Learning Network Model

In order to solve the problem that individual coordinates are easily ignored in the localization of abnormal behavior of marine fish, resulting in low recognition accuracy, execution efficiency and high false alarm rate, this paper proposes a method of fish abnormal behavior recognition based on deep learning network model. Firstly, the shadow of the fish behavior data is removed, and the background image is subtracted from each frame image to get the gray image of the fish school. Then, the label watershed algorithm is used to identify the fish, and the coordinates of different individuals in the fish swarm are obtained. Combined with the experimental size constraints and the number of fish, and combined with the deep learning network model, the weak link of video tag monitoring of abnormal behavior of marine fish is analyzed. Finally, the multi instance learning method and dual flow network model are used to identify the anomaly of marine fish school. The experimental results show that the method has high recognition accuracy, low false alarm rate and high execution efficiency. This method can provide a practical reference for the related research in this field.

A physics-informed deep learning framework for inversion and surrogate modeling in solid mechanics

We present the application of a class of deep learning, known as Physics Informed Neural Networks (PINN), to inversion and surrogate modeling in solid mechanics. We explain how to incorporate the momentum balance and constitutive relations into PINN, and explore in detail the application to linear elasticity, and illustrate its extension to nonlinear problems through an example that showcases von Mises elastoplasticity. While common PINN algorithms are based on training one deep neural network (DNN), we propose a multi-network model that results in more accurate representation of the field variables. To validate the model, we test the framework on synthetic data generated from analytical and numerical reference solutions. We study convergence of the PINN model, and show that Isogeometric Analysis (IGA) results in superior accuracy and convergence characteristics compared with classic low-order Finite Element Method (FEM). We also show the applicability of the framework for transfer learning, and find vastly accelerated convergence during network re-training. Finally, we find that honoring the physics leads to improved robustness: when trained only on a few parameters, we find that the PINN model can accurately predict the solution for a wide range of parameters new to the network—thus pointing to an important application of this framework to sensitivity analysis and surrogate modeling.

Towards Automatic Construction of Multi-Network Models for Heterogeneous Multi-Task Learning

Multi-task learning, as it is understood nowadays, consists of using one single model to carry out several similar tasks. From classifying hand-written characters of different alphabets to figuring out how to play several Atari games using reinforcement learning, multi-task models have been able to widen their performance range across different tasks, although these tasks are usually of a similar nature. In this work, we attempt to expand this range even further, by including heterogeneous tasks in a single learning procedure. To do so, we firstly formally define a multi-network model, identifying the necessary components and characteristics to allow different adaptations of said model depending on the tasks it is required to fulfill. Secondly, employing the formal definition as a starting point, we develop an illustrative model example consisting of three different tasks (classification, regression, and data sampling). The performance of this illustrative model is then analyzed, showing its capabilities. Motivated by the results of the analysis, we enumerate a set of open challenges and future research lines over which the full potential of the proposed model definition can be exploited.

A Multi-Cooperative Deep Convolutional Neural Network for Spatiotemporal Satellite Image Fusion

Remote sensing satellite images with high temporal and high spatial resolution play a critical role in earth science applications. However, it is difficult for a single satellite to obtain such images due to technical and cost constraints. Therefore, spatiotemporal image fusion based on deep learning has received extensive attention in recent years. This article proposes a multicooperative deep convolutional neural network (MCDNet) for spatiotemporal satellite image fusion. This method is a new multinetwork model in which multiple networks work together to reconstruct the predicted image. The multinetwork model consists of a super-resolution network, a difference reconstruction network, and a collaborative training network. First, the super-resolution network uses the combination of a novel multiscale mechanism and dilated convolutions to make full use of the spectral information of the coarse image and upgrade it to a transitional image that matches the fine image. The difference reconstruction network uses structural relevance to complete the reconstruction of the fine difference image. The collaborative training network extracts the hidden information from the fine image and uses the time relevance to restrict the training of the difference reconstruction network. Finally, the fine difference image and the known fine image are combined to complete the image fusion. The new compound loss function can help multinetwork models better complete cooperative training. Through experiments on two datasets and comparison with existing fusion algorithms, the subjective and objective results prove that MCDNet can effectively reconstruct higher-quality prediction images.