Network Architecture Research Articles

Optimal and scalable entanglement distribution over crossbar quantum networks.

Crossbar networks are a cornerstone of network architectures, capable of operating both as standalone interconnections or as integral switching components in complex, multi-stage systems. The main advantages of crossbar networks are their non-blocking operation and unparalleled minimal latency. With the advent of large scale quantum networks, crossbars might be an important asset towards the Quantum Internet. This study proposes a solution for the problem of distributing entanglement within crossbar quantum networks. Entangled particles are a consumable resource in quantum networks, and are being used by most quantum protocols. By ensuring that nodes within quantum networks are being supplied with entanglement, the reliability and efficiency of the network is maintained. By providing an efficient, scalable framework that can be used to achieve optimal entanglement distribution within crossbar quantum networks, this study offers a theoretical achievement which can be also used for enhancing quantum network performance. An algorithm for selecting an optimal entanglement distribution configuration is proposed and fully tested on realistic possible configurations.

A new XR-based human‐robot collaboration assembly system based on industrial metaverse

The industrial metaverse is an emerging paradigm of integrated information technologies. A human-robot collaboration assembly system is developed using extended reality (XR) and blockchain technologies. This system includes the physical, virtual reality (VR), augmented reality (AR) supporting, and blockchain systems. Virtual avatars represent physical operators and mobile robots in the XR environment. The VR system emulates real-time physical operations, facilitating pre-planning, online task optimization, and converting strategic decisions into executable instructions for physical robots. The AR system enhances the cognitive experience of intelligent operators, contributing to decision-making and VR system control through multi-modal interactions. The integration of the VR and AR systems in perception-decision-control enables real-time process optimization and the reverse control of physical operations. The system is implemented using a decentralized blockchain and end-edge-cloud collaborative network architecture to regulate interaction permissions for each scene characters, ensuring systematic functionality, robust communication, and comprehensive data management. A case study of a gearbox assembly is used to demonstrate the system’s development and efficacy. The results indicate the VR system's comprehensive perception and the AR system's detailed insights, yielding reliable information on global and local levels. The integration of artificial intelligence (AI) algorithms and spatiotemporal visualization during VR decision-making provides computational advantages. Incorporating human input into the AR system enables human-in-the-loop operational decision-making, providing optimal instructions to improve human-robot collaboration performance. Notably, the fusion of blockchain and end-edge-cloud architectures augments inter-system data exchange by efficiently managing and processing human-robot interactions and their diverse avatars. This system offers an innovative approach to human-robot collaboration assembly in manufacturing, with implications for analogous contexts.

Application of Convolutional Networks for Localization and Prediction of Scalar Parameters of Fractured Geological Inclusion

Seismic inversion is an important part of the modern geological exploration process. Novel applications of deep learning are capable of handling heterogeneous media, but require too much data for training. In this paper, we focus on the prediction of fracture inclusion location and its parameters in rock media and approach the problem in the multi-task manner. For this, several multi-task convolutional neural network (CNN) architectures are proposed and compared. The direct seismic problem is considered in the heterogeneous fractured geological model based on the well-known Marmousi2 model in a two-dimensional case. The model of the deformable solid medium containing slip planes with nonlinear slip conditions on them and explicit–implicit numerical method is applied to obtain the synthetic seismic dataset for CNN training and validation.

A Hybrid Learning-Architecture for Improved Brain Tumor Recognition

The accurate classification of brain tumors is an important step for early intervention. Artificial intelligence (AI)-based diagnostic systems have been utilized in recent years to help automate the process and provide more objective and faster diagnosis. This work introduces an enhanced AI-based architecture for improved brain tumor classification. We introduce a hybrid architecture that integrates vision transformer (ViT) and deep neural networks to create an ensemble classifier, resulting in a more robust brain tumor classification framework. The analysis pipeline begins with preprocessing and data normalization, followed by extracting three types of MRI-derived information-rich features. The latter included higher-order texture and structural feature sets to harness the spatial interactions between image intensities, which were derived using Haralick features and local binary patterns. Additionally, local deeper features of the brain images are extracted using an optimized convolutional neural networks (CNN) architecture. Finally, ViT-derived features are also integrated due to their ability to handle dependencies across larger distances while being less sensitive to data augmentation. The extracted features are then weighted, fused, and fed to a machine learning classifier for the final classification of brain MRIs. The proposed weighted ensemble architecture has been evaluated on publicly available and locally collected brain MRIs of four classes using various metrics. The results showed that leveraging the benefits of individual components of the proposed architecture leads to improved performance using ablation studies.

A novel CNN-based image segmentation pipeline for individualized feline spinal cord stimulation modeling.

Spinal cord stimulation (SCS) is a well-established treatment for the management of certain chronic pain conditions. More recently, it has also garnered attention as a means of modulating neural activity with the goal of restoring lost autonomic or sensory-motor function. Personalized modeling and treatment planning are critical aspects of safe and effective SCS [46, 60]. However, the generation of spine models at the required level of detail and accuracy requires time and labor intensive manual image segmentation by human experts. Hence, there is a need for maximally automated segmentation routines capable of producing high-quality anatomical models that can be used even in cases where available data is limited. To this end, we developed an automated image segmentation and model generation pipeline based on a novel Convolutional Neural Network (CNN) architecture trained on feline spinal cord magnetic resonance imaging (MRI) data. The pipeline includes steps for image preprocessing, data augmentation, transfer learning and cleanup. To assess the relative importance of each step in the pipeline and of our choice of CNN architecture, we systematically dropped steps or substituted architectures, quantifying the downstream effects in terms of tissue segmentation quality (Jaccard index and Hausdorff distance) and predicted nerve recruitment (estimated axonal depolarization). This leaveone-out analysis demonstrated that each pipeline step contributed a small but measurable increment to mean segmentation quality. Surprisingly, minor differences in segmentation accuracy translated to significant deviations (ranging between 4% and 13% for each pipeline step) in predicted nerve recruitment, highlighting the importance of careful workflow design. To our knowledge, this is the first analysis to also assess the downstream impact of segmentation quality differences on neurostimulation predictions. Furthermore, transfer learning techniques enhanced segmentation metric consistency and allowed generalization to a completely different spine region with minimal additional training data. This work helps pave the way towards fully automated, personalized SCS treatment planning in clinical settings.

Mobile Application Based on Convolutional Neural Networks for Pterygium Detection in Anterior Segment Eye Images at Ophthalmological Medical Centers

This article introduces an innovative mobile solution for Pterygium detection, an eye disease, using a classification model based on the convolutional neural network (CNN) architecture ResNext50 in images of the anterior segment of the eye. Four models (ResNext50, ResNet50, MobileNet v2, and DenseNet201) were used for the analysis, with ResNext50 standing out for its high accuracy and diagnostic efficiency. The research, focused on applications for ophthalmological medical centers in Lima, Peru, explains the process of development and integration of the ResNext50 model into a mobile application. The results indicate the high effectiveness of the system, highlighting its high precision, recall, and specificity, which exceed 85%, thus showing its potential as an advanced diagnostic tool in ophthalmology. This system represents a significant tool in ophthalmology, especially for areas with limited access to specialists, offering a rapid and reliable diagnosis of Pterygium. The study also addresses the technical challenges and clinical implications of implementing this technology in a real-world context.

Unifying O(3) equivariant neural networks design with tensor-network formalism

Many learning tasks, including learning potential energy surfaces from ab initio calculations, involve global spatial symmetries and permutational symmetry between atoms or general particles. Equivariant graph neural networks are a standard approach to such problems, with one of the most successful methods employing tensor products between various tensors that transform under the spatial group. However, as the number of different tensors and the complexity of relationships between them increase, maintaining parsimony and equivariance becomes increasingly challenging. In this paper, we propose using fusion diagrams, a technique widely employed in simulating SU(2)-symmetric quantum many-body problems, to design new spatial equivariant components for neural networks. This results in a diagrammatic approach to constructing novel neural network architectures. When applied to particles within a given local neighborhood, the resulting components, which we term ‘fusion blocks,’ serve as universal approximators of any continuous equivariant function defined on the neighborhood. We incorporate a fusion block into pre-existing equivariant architectures (Cormorant and MACE), leading to improved performance with fewer parameters on a range of challenging chemical problems. Furthermore, we apply group-equivariant neural networks to study non-adiabatic molecular dynamics of stilbene cis-trans isomerization. Our approach, which combines tensor networks with equivariant neural networks, suggests a potentially fruitful direction for designing more expressive equivariant neural networks.

Improved Galaxy Morphology Classification with Convolutional Neural Networks

The increased volume of images and galaxies surveyed by recent and upcoming projects consolidates the need for accurate and scalable automated AI-driven classification methods. This paper proposes a new algorithm based on a custom neural network architecture for classifying galaxies from deep space surveys. The convolutional neural network (CNN) presented is trained using 10,000 galaxy images obtained from the Galaxy Zoo 2 dataset. It is designed to categorize galaxies into five distinct classes: completely round smooth, in-between smooth (falling between completely round and cigar-shaped), cigar-shaped smooth, edge-on, and spiral. The performance of the proposed CNN is assessed using a set of metrics such as accuracy, precision, recall, F1 score, and area under the curve. We compare our solution with well-known architectures like ResNet-50, DenseNet, EfficientNet, Inception, MobileNet, and one proposed model for galaxy classification found in the recent literature. The results show an accuracy rate of 96.83%, outperforming existing algorithms.

Strategies for multi-case physics-informed neural networks for tube flows: a study using 2D flow scenarios

Fluid dynamics computations for tube-like geometries are crucial in biomedical evaluations of vascular and airways fluid dynamics. Physics-Informed Neural Networks (PINNs) have emerged as a promising alternative to traditional computational fluid dynamics (CFD) methods. However, vanilla PINNs often demand longer training times than conventional CFD methods for each specific flow scenario, limiting their widespread use. To address this, multi-case PINN approach has been proposed, where varied geometry cases are parameterized and pre-trained on the PINN. This allows for quick generation of flow results in unseen geometries. In this study, we compare three network architectures to optimize the multi-case PINN through experiments on a series of idealized 2D stenotic tube flows. The evaluated architectures include the ‘Mixed Network’, treating case parameters as additional dimensions in the vanilla PINN architecture; the “Hypernetwork”, incorporating case parameters into a side network that computes weights in the main PINN network; and the “Modes” network, where case parameters input into a side network contribute to the final output via an inner product, similar to DeepONet. Results confirm the viability of the multi-case parametric PINN approach, with the Modes network exhibiting superior performance in terms of accuracy, convergence efficiency, and computational speed. To further enhance the multi-case PINN, we explored two strategies. First, incorporating coordinate parameters relevant to tube geometry, such as distance to wall and centerline distance, as inputs to PINN, significantly enhanced accuracy and reduced computational burden. Second, the addition of extra loss terms, enforcing zero derivatives of existing physics constraints in the PINN (similar to gPINN), improved the performance of the Mixed Network and Hypernetwork, but not that of the Modes network. In conclusion, our work identified strategies crucial for future scaling up to 3D, wider geometry ranges, and additional flow conditions, ultimately aiming towards clinical utility.

A three-dimensional, discrete-continuum model of blood pressure in microvascular networks.

We present a 3D discrete-continuum model to simulate blood pressure in large microvascular tissues in the absence of known capillary network architecture. Our hybrid approach combines a 1D Poiseuille flow description for large, discrete arteriolar and venular networks coupled to a continuum-based Darcy model, point sources of flux, for transport in the capillary bed. We evaluate our hybrid approach using a vascular network imaged from the mouse brain medulla/pons using multi-fluorescence high-resolution episcopic microscopy (MF-HREM). We use the fully-resolved vascular network to predict the hydraulic conductivity of the capillary network and generate a fully-discrete pressure solution to benchmark against. Our results demonstrate that the discrete-continuum methodology is a computationally feasible and effective tool for predicting blood pressure in real-world microvascular tissues when capillary microvessels are poorly defined.

Network architecture for single image super‐resolution: A comprehensive review and comparison

AbstractSingle image super‐resolution (SISR) is a promising research direction in computer vision and image processing for improving the visual perception of low‐quality images. In recent years, deep learning algorithms have driven tremendous development in SR, and SR methods based on various network architectures have significantly improved the quality of reconstructed images. Although there has been a large amount of reviews focusing on SISR, few studies have focused specifically on network architectures for SISR. This paper aims to provide a systematic overview of the design ideas of SISR using multiple architectures, including Convolutional Neural Networks (CNN), Generative Adversarial Networks (GAN), Transformer, and Diffusion model. In addition, an experimental analysis and comparison of state‐of‐the‐art SR algorithms have been performed on publicly available quantitative and qualitative datasets. Finally, some future directions are discussed that may help other community researchers.

Comparative Analysis of CNN Architectures for SIBI Image Classification

The classification of images from the Indonesian Sign Language System (SIBI) using VGG16, ResNet50, Inception, Xception, and MobileNetV2 Convolutional Neural Network (CNN) architectures is evaluated in this paper. With Google Colab Pro, a 224 × 224-pixel picture dataset was used for the study. A five-stage technique consisting of Dataset Collection, Dataset Preprocessing, Model Design, Model Training, and Model Testing was applied. Performance evaluation focused on accuracy, precision, recall, and F1-Score. The results identified VGG16 as the top-performing model with an accuracy of 99.60% and an equivalent F1-Score, followed closely by ResNet50 with nearly similar performance. Inception, XCeption, and MobileNetV2 demonstrated balanced performance but with lower accuracy. This study sheds light on the best CNN models to choose for SIBI image classification, and it makes recommendations for further research that include using sophisticated data augmentation methods, investigating novel CNN architectures, and putting the models to practical use.

A Lightweight Face Detector via Bi-Stream Convolutional Neural Network and Vision Transformer

Lightweight convolutional neural networks are widely used for face detection due to their ability to learn local representations through spatial induction bias and translational invariance. However, convolutional face detectors have limitations in detecting faces under challenging conditions like occlusion, blurring, or changes in facial poses, primarily attributed to fixed-size receptive fields and a lack of global modeling. Transformer-based models have advantages on learning global representations but are insensitive to capture local patterns. To address these limitations, we propose an efficient face detector that combines convolutional neural network and transformer architectures. We introduce a bi-stream structure that integrates convolutional neural network and transformer blocks within the backbone network, enabling the preservation of local pattern features and the extraction of global context. To further preserve the local details captured by convolutional neural networks, we propose a feature enhancement convolution block in a hierarchical backbone structure. Additionally, we devise a multiscale feature aggregation module to enhance obscured and blurred facial features. Experimental results demonstrate that our method has achieved improved lightweight face detection accuracy with an average precision of 95.30%, 94.20%, and 87.56% across the easy, medium, and hard subdatasets of WIDER FACE, respectively. Therefore, we believe our method will be a useful supplement to the collection of current artificial intelligence models and benefit the engineering applications of face detection.

A One-Step Methodology for Identifying Concrete Pathologies Using Neural Networks—Using YOLO v8 and Dataset Review

Pathologies in concrete structures can be visually evidenced on the concrete surface, such as by fissures or cracks, fragmentation of part of the concrete, concrete efflorescence, corrosion stains on the concrete surface, or exposed steel bars, the latter two occurring in reinforced concrete. Therefore, these pathologies can be analyzed via the images of concrete structures. This article proposes a methodology for visually inspecting concrete structures using deep neural networks. This method makes it possible to speed up the detection task and increase its effectiveness by saving time in preparing the identifications to be analyzed and eliminating or reducing errors, such as those resulting from human errors caused by the execution of tedious, repetitive analysis tasks. The methodology was tested to analyze its accuracy. The neural network architecture used for detection was YOLO, versions 4 and 8, which was tested to analyze the gain with migration to a more recent version. The dataset for classification was Ozgnel, which was trained with YOLO version 8, and the detection dataset was CODEBRIM. The use of a dedicated classification dataset allows for a better-trained network for this function and results in the elimination of false positives in the detection stage. The classification achieved 99.65% accuracy.

SYNTHETIC DATA GENERATION FOR ANN MODELING OF THE HYDRODYNAMIC PROCESSES OF IN-SITU LEACHING

The work presents an approach to enhance the forecasting capabilities of In-Situ Leaching processes during both the production stage and early prognosis. ISL, a crucial method for resource extraction, demands rapid on-site forecasting to guide the deployment of new technological blocks. Traditional modeling techniques, though effective, are hindered by their computational demands and network throughput requirements, particularly when dealing with substantial datasets or remote computing needs. The integration of AI technologies, specifically neural networks, offers a promising opportunity for expedited calculations by leveraging the power of forward propagation through pretrained neural models. However, a critical challenge lies in transforming conventional numerical datasets into a format suitable for neural modeling. Furthermore, the scarcity of training data during the production phase, where vital parameters are concealed underground, poses an additional challenge in training AI models for In-Situ Leaching processes. This research addresses these challenges by proposing a methodology for generating training data tailored to the most resource-intensive Computational Fluid Dynamics problems encountered during modeling. Traditional numerical modeling techniques are harnessed to construct training datasets comprising input and corresponding expected output data, with a particular focus on varying well network patterns. Subsequent efforts are directed at the conversion of the acquired data into a format compatible with neural networks. The data is normalized to align with the data ranges stipulated by the activation functions employed within the neural network architecture. This preprocessing step ensures that the neural model can effectively learn from the generated data, facilitating accurate forecasting of In-Situ Leaching processes. An advantage of proposed technique lies in provision of large, reliable datasets to train neural network to predict hydrodynamic properties based on technological regimes currently active or expected on ISL site. A major implication of this approach lies in applicability of pre-trained AI technologies to forecast future or determine current hydrodynamic regime in the stratum circumventing cost deterministic simulations currently deployed at mining sites. Hence, innovative approach outlined in this paper holds promise for optimizing forecasting, allowing for quicker and more efficient decision-making in resource extraction operations while getting around the computational barriers associated with traditional methods.

Sampling the user controls in neural modeling of audio devices

This work studies neural modeling of nonlinear parametric audio circuits, focusing on how the diversity of settings of the target device user controls seen during training affects network generalization. To study the problem, a large corpus of training datasets is synthetically generated using SPICE simulations of two distinct devices, an analog equalizer and an analog distortion pedal. A proven recurrent neural network architecture is trained using each dataset. The difference in the datasets is in the sampling resolution of the device user controls and in their overall size. Based on objective and subjective evaluation of the trained models, a sampling resolution of five for the device parameters is found to be sufficient to capture the behavior of the target systems for the types of devices considered during the study. This result is desirable, since a dense sampling grid can be impractical to realize in the general case when no automated way of setting the device parameters is available, while collecting large amounts of data using a sparse grid only incurs small additional costs. Thus, the result provides guidance for efficient collection of training data for neural modeling of other similar audio devices.

Classification of Various Plant Leaf Disease Using Pretrained Convolutional Neural Network On Imagenet

Introduction/Background Plant diseases and pernicious insects are a considerable threat in the agriculture sector. Leaf diseases impact agricultural production. Therefore, early detection and diagnosis of these diseases are essential. This issue can be addressed if a farmer can detect the diseases properly. Objective The fundamental goal of this project is to create and test a model for precisely classifying leaf diseases in plants. Materials and Methods This paper introduces a model designed to classify leaf diseases effectively. The research utilizes the publicly available PlantVillage dataset, which includes 38 different classes of leaf images, ranging from healthy to disease-infected leaves. Pretrained CNN (Convolutional Neural Network) models, including VGG16, ResNet50, InceptionV3, MobileNetV2, AlexNet, and EfficientNet, are employed for image classification. Results The paper provides a performance comparison of these models. The results show that the EfficientNet model achieves an accuracy of 97.5% in classifying healthy and diseased leaf images, outperforming other models. Discussion This research highlights the potential of utilizing advanced neural network architectures for accurate disease detection in the agricultural sector. Conclusion This study demonstrates the efficacy of employing sophisticated CNN models, particularly EfficientNet, to properly identify leaf diseases. Such technological developments have the potential to improve disease detection in agriculture. These improvements help to improve food security by allowing for preventive actions to battle crop diseases.

Accelerating handwritten number classifier with pipelined Multiply-Accumulate units in neural networks

This research project explores the development and assessment of a Handwritten Number Classifier based on Neural Networks, employing the widely used MNIST dataset. The primary objective of this study is to enhance computational efficiency, and to achieve this, it concentrates on integrating and contrasting non-pipelined and pipelined structures within the Multiply-Accumulate (MAC) unit. The initial phase involved conducting MATLAB simulations, which yielded promising results regarding the accuracy of weight calculations. Subsequently, Hardware Description Language (HDL) testing was carried out to further validate the classifiers performance. In the HDL testing phase, the classifier incorporating the pipelined MAC unit demonstrated a substantial 42.9% enhancement in processing speed when compared to its non-pipelined counterpart. These results highlight the potential advantages of employing pipeline processing in neural network architectures, emphasizing its effectiveness in achieving faster and more efficient image classification, particularly when dealing with extensive datasets. In conclusion, this research project not only presents valuable insights into improving the efficiency of neural network-based image classifiers but also lays the groundwork for potential future endeavors. These future directions may include adapting the classifier to handle more complex datasets and addressing emerging challenges.

Connections between soil microbes, land use and European climate: Insights for management practices

Soil microbial biomass and activity strongly depend on land use, vegetation cover, climate, and soil physicochemical properties. In most cases, this dependence was assessed by one-to-one correlations while by employing network analysis, information about network robustness and the balance between stochasticity and determinism controlling connectivity, was revealed. In this study, we further elaborated on the hypothesis of Smith et al. (2021) that cropland soils depended more on climate variables and therefore are more vulnerable to climate change. We used the same dataset with that of Smith et al. (2021) that contains seasonal microbial, climate and soil variables collected from 881 soil points representing the main land uses in Europe: forests, grassland, cropland. We examined complete (both direct and indirect relationships) and incomplete networks (only direct relationships) and recorded higher robustness in the former. Partial Least Square results showed that on average more than 45% of microbial attributes' variability was predicted by climate and habitat drivers denoting medium to strong effect of habitat filtering. Network architecture slightly affected by season or land use type; it followed the core/periphery structure with positive and negative interactions and no hub nodes. Microbial attributes (biomass, activity and their ratio) mostly belong to core block together with Soil Organic Carbon (SOC), while climate and soil variables to periphery block with the exception of cropland networks, denoting the higher dependence between microbial and climate variables in these latter. All complete networks appeared robust except for cropland and forest in summer, a finding that disagrees with our initial hypothesis about cropland. Networks' connectivity was controlled stronger by stochasticity in forest than in croplands. The lack of human interventions in forest soils increase habitat homogeneity enhancing the influence of stochastic agents such as microbial unlimited dispersal and/or stochastic extinction. The increased stochasticity implies the necessity for proactive management actions.

A Multiscale Parallel Pedestrian Recognition Algorithm Based on YOLOv5

Mainstream pedestrian recognition algorithms have problems such as low accuracy and insufficient real-time performance. In this study, we developed an improved pedestrian recognition algorithm named YOLO-MSP (multiscale parallel) based on residual network ideas, and we improved the network architecture based on YOLOv5s. Three pooling layers were used in parallel in the MSP module to output multiscale features and improve the accuracy of the model while ensuring real-time performance. The Swin Transformer module was also introduced into the network, which improved the efficiency of the model in image processing by avoiding global calculations. The CBAM (Convolutional Block Attention Module) attention mechanism was added to the C3 module, and this new module was named the CBAMC3 module, which improved model efficiency while ensuring the model was lightweight. The WMD-IOU (weighted multidimensional IOU) loss function proposed in this study used the shape change between the recognition frame and the real frame as a parameter to calculate the loss of the recognition frame shape, which could guide the model to better learn the shape and size of the target and optimize recognition performance. Comparative experiments using the INRIA public data set showed that the proposed YOLO-MSP algorithm outperformed state-of-the-art pedestrian recognition methods in accuracy and speed.