Redundant Network Research Articles

Optimizing Systemic Redundancy of Traffic Sensor Networks While Maintaining Resilience: New Evidence From Using Graph Learning

The optimization of systemic redundancy by minimizing the sensor quantity can improve the efficiency of sensor networks and save costs. However, from the perspective of risk management, this redundancy reduction can also bring a significant loss in the overall network resilience because the less the systemic redundancy is, the fewer backup components in the network when shocks hit and, therefore, the less overall resilience. In this article, we investigate this intractable dilemma and attempt to pinpoint the tradeoff point for a city-scale automatic number plate recognition (ANPR) system in Cambridge, U.K. By developing a two-stage graph deep learning model, we first optimize the layout of the ANPR system to reduce redundancy and find its efficiency profile. Next, we study what effects this redundancy reduction can bring to the overall resilience, as the overall observability drops with the reduction in the number of sensors and find an optimal balance. The results show that our approach can effectively optimize the system's redundancy by using only 47% of the original sensors to reconstruct the full picture with a mean absolute error of only 11.18 and a root mean square error of 19.49; most importantly, the overall system resilience is maintained at 70% in the meantime. This article provides an alternative perspective for dealing with the well-known “efficiency-resilience” dilemma and offers new evidence to enable better decision and policy making for city managers and planners in local authorities.

YOLO-Drone: An Optimized YOLOv8 Network for Tiny UAV Object Detection

With the widespread use of UAVs in commercial and industrial applications, UAV detection is receiving increasing attention in areas such as public safety. As a result, object detection techniques for UAVs are also developing rapidly. However, the small size of drones, complex airspace backgrounds, and changing light conditions still pose significant challenges for research in this area. Based on the above problems, this paper proposes a tiny UAV detection method based on the optimized YOLOv8. First, in the detection head component, a high-resolution detection head is added to improve the device’s detection capability for small targets, while the large target detection head and redundant network layers are cut off to effectively reduce the number of network parameters and improve the detection speed of UAV; second, in the feature extraction stage, SPD-Conv is used to extract multi-scale features instead of Conv to reduce the loss of fine-grained information and enhance the model’s feature extraction capability for small targets. Finally, the GAM attention mechanism is introduced in the neck to enhance the model’s fusion of target features and improve the model’s overall performance in detecting UAVs. Relative to the baseline model, our method improves performance by 11.9%, 15.2%, and 9% in terms of P (precision), R (recall), and mAP (mean average precision), respectively. Meanwhile, it reduces the number of parameters and model size by 59.9% and 57.9%, respectively. In addition, our method demonstrates clear advantages in comparison experiments and self-built dataset experiments and is more suitable for engineering deployment and the practical applications of UAV object detection systems.

Designing an automatic pollen monitoring network for direct usage of observations to reconstruct the concentration fields

We consider several approaches to a design of a regional-to-continent-scale automatic pollen monitoring network in Europe. Practical challenges related to the arrangement of such a network limit the range of possible solutions. A hierarchical network is discussed, highlighting the necessity of a few reference sites that follow an extended observations protocol and have corresponding capabilities.Several theoretically rigorous approaches to a network design have been developed so far. However, before starting the process, a network purpose, a criterion of its performance, and a concept of the data usage should be formalized. For atmospheric composition monitoring, developments follow one of the two concepts: a network for direct representation of concentration fields and a network for model-based data assimilation, inverse problem solution, and forecasting. The current paper demonstrates the first approach, whereas the inverse problems are considered in a follow-up paper.We discuss the approaches for the network design from theoretical and practical standpoints, formulate criteria for the network optimality, and consider practical constraints for an automatic pollen network. An application of the methodology is demonstrated for a prominent example of Germany's pollen monitoring network. The multi-step method includes (i) the network representativeness and (ii) redundancy evaluation followed by (iii) fidelity evaluation and improvement using synthetic data.

Toward Resilient Network Slicing for Satellite–Terrestrial Edge Computing IoT

Satellite-Terrestrial Edge Computing Networks (STECNs) emerged as a global solution to support multiple Internet of Things (IoT) applications in 6G networks. The enabling technologies to slice STECNs such as Software-Defined Networking (SDN), satellite edge computing, and Network Function Virtualization (NFV) are key to realizing this vision. In this paper, we survey and analyze network slicing solutions for STECNs. We discuss slice management and orchestration for different STECNs integration architectures, satellite edge computing, mmWave/THz, and AI solutions to make network slicing adaptive. In addition, we identify challenges and open issues to slice STECNs. In particular, resilient network slicing is crucial for essential and critical services. Network failures are unavoidable in large networks and can cause significant disruptions in network slicing, compromising many services. To this end, we present a resilient network slicing design to cope with failures and guarantee service continuity which is agnostic to the integration architecture and inherently multi-domain. Further, we present strategies to achieve resilient networking and slicing in STECNs including planning and provisioning of redundant network resources, design rules for service level agreement decomposition, and cross-domain solutions to detect and mitigate failures. Finally, promising future research directions are highlighted. This paper provides valuable guidelines for slicing STECNs and will benefit key sectors, such as smart healthcare, e-commerce, industrial IoT, and education, among others.

Climate change–induced stress disrupts ectomycorrhizal interaction networks at the boreal–temperate ecotone

The interaction networks formed by ectomycorrhizal fungi (EMF) and their tree hosts, which are important to both forest recruitment and ecosystem carbon and nutrient retention, may be particularly susceptible to climate change at the boreal-temperate forest ecotone where environmental conditions are changing rapidly. Here, we quantified the compositional and functional trait responses of EMF communities and their interaction networks with two boreal (Pinus banksiana and Betula papyrifera) and two temperate (Pinus strobus and Quercus macrocarpa) hosts to a factorial combination of experimentally elevated temperatures and reduced rainfall in a long-term open-air field experiment. The study was conducted at the B4WarmED (Boreal Forest Warming at an Ecotone in Danger) experiment in Minnesota, USA, where infrared lamps and buried heating cables elevate temperatures (ambient, +3.1 °C) and rain-out shelters reduce growing season precipitation (ambient, ~30% reduction). EMF communities were characterized and interaction networks inferred from metabarcoding of fungal-colonized root tips. Warming and rainfall reduction significantly altered EMF community composition, leading to an increase in the relative abundance of EMF with contact-short distance exploration types. These compositional changes, which likely limited the capacity for mycelial connections between trees, corresponded with shifts from highly redundant EMF interaction networks under ambient conditions to less redundant (more specialized) networks. Further, the observed changes in EMF communities and interaction networks were correlated with changes in soil moisture and host photosynthesis. Collectively, these results indicate that the projected changes in climate will likely lead to significant shifts in the traits, structure, and integrity of EMF communities as well as their interaction networks in forest ecosystems at the boreal-temperate ecotone.

Economic evaluation of redundancy design for transportation networks under disruptions: Framework and case study

Adding transportation network redundancy is an effective strategy to hedge against disruptions in the pre-disaster planning stage. However, given the low frequency of disruptions and large financial cost of redundancy design, the economic feasibility of redundancy design has always been challenged. This paper develops a framework to investigate the economic feasibility of adding redundancy to road networks in the pre-disaster planning stage by considering network-wide benefits and stochastic disruption scenarios. Specifically, the redundancy of road networks is defined as route diversity, which refers to the number of behaviorally alternative routes between each origin-destination pair. The redundancy design schemes are formulated by adding parallel links to critical links in current road networks, as these added links provide alternative routes for numerous travelers when the critical links are disrupted. As the redundancy of a transportation system is not only accessible under disruptions but also on duty in normal states, this paper estimates the benefits of redundancy design for road networks in both normal and disruptive states. The Winnipeg network is then used in the case study with the focus on adding redundant bridges. Results indicate that the benefit-cost ratio (BCR) of adding one redundant bridge is 3.51 and redundancy design in road networks is economically feasible. Since the value of BCR is sensitive to the discount rate and the performance of disruptive events, the related data should be collected cautiously when estimating the benefits of redundancy design in practice. Meanwhile, we observe the diminishing marginal effect in the addition of multiple redundant links, which provides insights into the decision-making regarding the optimal amount of redundancy investment in road networks. The proposed framework can be considered as a decision support tool to justify and rank alternative projects from the new perspective of redundancy and resilience.

The effects of fusion-based feature extraction for fabric defect classification

Image processing has been employed in a variety of fields since the advent of image processing techniques. One of these fields is textiles. The existence of any defect in a fabric is one of the most important factors affecting the quality of the fabric. There are many types of fabric defects that can occur for various reasons. It is critical to figure out what caused the defect and fix it so that it does not occur again. Automation of fabric defect detection has recently attracted a great deal of interest in view of the development in artificial intelligence technology to be able to discover defects with a high degree of success and to limit the harm to the manufacturer. This study focuses on analyzing different feature extraction methods and different classifiers and discussing the advantages/disadvantages of the combinations and, unlike other studies, using feature fusion for feature extraction. Different cases have been created that handle fabric datasets from different angles and apply different methods of feature extraction (convolution neural network, minimum relevance and maximum redundancy) and classification (ensemble learning (EL), k-nearest neighbor, support vector machine (SVM)) for separating defected and un-defected patterned and un-patterned fabrics. ResNet18 is the convolution neural network-based model with the highest performance in feature extraction, while EL and the SVM allow us to achieve close and highly successful results in classification. When feature fusion is used, ResNet18 & GoogLeNet & SVM is the most successful combination compared to the others (94.66%).

Neural-network-based parameter tuning for multi-agent simulation using deep reinforcement learning

This study proposes a new efficient parameter tuning method for multi-agent simulation (MAS) using deep reinforcement learning. MAS is currently a useful tool for social sciences, but is hard to realize realistic simulations due to its computational burden for parameter tuning. This study proposes efficient parameter tuning to address this issue using deep reinforcement learning methods. To improve compatibility with the tuning task, our proposed method employs actor-critic-based deep reinforcement learning, such as deep deterministic policy gradient (DDPG) and soft actor-critic (SAC). In addition to the customized version of DDPG and SAC for our task, we also propose three additional components to stabilize the learning: an action converter (DDPG only), a redundant full neural network actor, and a seed fixer. For experimental verification, we employ a parameter tuning task in an artificial financial market simulation, comparing our proposed model, its ablations, and the Bayesian estimation-based baseline. The results demonstrate that our model outperforms the baseline in terms of tuning performance, indicating that the additional components of the proposed method are essential. Moreover, the critic of our model works effectively as a surrogate model, that is, as an approximate function of the simulation, which allows the actor to tune the parameters appropriately. We have also found that the SAC-based method exhibits the best and fastest convergence, which we assume is achieved by the high exploration capability of SAC.

Exploratory Analysis of Surrogate Metrics to Assess the Resilience of Water Distribution Networks

AbstractThis study compares and discusses the adequacy of surrogate resilience metrics proposed in the literature for resilience assessment of drinking water systems concerning demand increase and network redundancy. A sensitivity analysis is carried out for increasing flow rates using a conceptual case study with different layouts and demand scenarios, selecting several metrics to assess the resilience of two real network areas. Resilience metrics based on surplus energy are sensitive to network layout and demand scenarios. The network resilience index considers hydraulic reliability and network diameter uniformity. In contrast, the weighted resilience index also considers the network topology and gives importance to pipes with higher flow rates. Entropy‐based resilience metrics mainly rely on the network flows' uniformity and are sensitive to pipe redundancy. The entropy metric most adequate to assess the hydraulic capacity is the diameter‐sensitive flow entropy, since it is sensitive to the velocity inside the pipes. Topology metrics cannot assess the hydraulic capacity though evaluate the system redundancy (meshed‐ness coefficient), robustness (central‐point dominance) and water transportation efficiency (average‐path length). Surrogate resilience metrics do not assess the system performance during a failure. They indicate systems which are better prepared to overcome failure events and increased demand events, providing vital information to drinking water systems management.

Target detection of substation electrical equipment from infrared images using an improved faster regions with convolutional neural network features algorithm

The failure of substation equipment can cause incalculable losses to the economy and power consumption of the whole country. The use of infrared images is a powerful tool to obtain equipment temperature, which can then be used directly to diagnose substation equipment without stopping the operation of the equipment. In this paper, the authors focus on the correct identification of different types of electrical equipment from the infrared images. An improved faster regions with convolutional neural network features (faster R-CNN) algorithm is proposed, which shows very high detection accuracy for substation equipment. Firstly, the backbone of the faster R-CNN is optimised. A new network, the ResNet-30 network, is designed to reduce the redundancy of the ResNet-34 network and increases the proportion of residual blocks in the network in the previous stages. Next, the deep receptive field is combined with the shallow receptive field of the network and a double-shortcut structure with a large convolutional kernel is proposed. This enhances the ability of network feature extraction. A cross-channel shortcut is proposed at the channel transition of the network based on the channel number relationship between the dual-shortcut structures. Finally, the proposed method is compared with faster R-CNNs whose backbones are ResNet-50 plus a feature pyramid network (ResNet-50+FPN), you only look once v3 plus spatial pyramid pooling (YOLOv3+SPP) and a single-shot multibox detector (SSD). The results show that the improved model not only has a smaller number of parameters and low requirements for graphics processing unit (GPU) equipment, but also has the highest mean average precision (mAP) for mostly substation equipment in the test-set. This lays a foundation for fault diagnosis of substation equipment in the future.

Regeneration of cytosolic thiol peroxidases.

Three soluble type two peroxiredoxins (PRXIIB, C, D) and two glutathione peroxidase-like enzymes (GPXL2, 8) reside in the cytosol of Arabidopsis thaliana cells and function both as thiol-dependent antioxidants and redox sensors. Their primary substrate is H2 O2 , but they also accept other peroxides with a distinct preference between PRXII and GPXL. Less known is their regeneration specificity in the light of the large set of thiol reductases, namely eight annotated thioredoxin h isoforms (TRXh1-5, 7-9), a few TRX-like proteins, including CxxS1 (formerly TRXh6) and several glutaredoxins (GRX) associated with the cytosol. This study addressed this open question by in vitro enzyme tests using recombinant protein. GPXL2 and 8 exclusively accepted electrons from the TRX system, namely TRXh1-5 and TDX, while PRXIIB/C/D were efficiently regenerated with GRXC1 and C2 but not the TRX-like protein Picot1. They showed significant but low activity (<3% of GRXC2) with TRXh1-5 and TDX. A similar reduction efficiency with TRX was seen in the insulin assay, only TDX was less active. Finally, the reduction of oxidized cytosolic malate dehydrogenase 1, as measured by regained activity, showed an extremely broad ability to accept electrons from different TRXs and GRXs. The results demonstrate redundancy and specificity in the redox regulatory network of the cytosol.

A fast linearized AC power flow-constrained robust unit commitment approach with customized redundant constraint identification method

The large-scale integration of renewable energy resources in the power system challenges its economic and secure operation. Particularly, the increasing penetration of renewable energy will result in insufficient system voltage regulation and reactive power support capabilities, and may cause high risks of nodal voltage and branch flow violations. Therefore, to hedge the operational risks under the worst realization of uncertainties of renewable energy sources, a two-stage robust unit commitment (UC) model is developed. Meanwhile, the convexified AC power flow model is incorporated in the robust UC model to more accurately characterize the real-time operating status of power systems. On this basis, an AC power flow-constrained robust unit commitment (ACRUC) model is formulated. A circular linearization method is then adopted to handle the quadratic constraints in the original AC power flow model, transforming them into tractable linear constraints. Furthermore, to reduce the computational complexity caused by the large-scale newly-added constraints after the linearization process, a customized redundant constraint identification (RCI) method is developed, in which two different modes (i.e., cold and warm start modes) are designed considering the difference in base case system operating condition for linearizing branch losses. Then, the redundant network security constraints could be identified by solving a series of relatively simple optimization subproblems. Numerical results on the modified NERL-118 test system indicate that the proposed model could accurately depict actual operation and scheduling conditions, and also verify that the proposed customized RCI method could effectively reduce the problem scale and improve the solution efficiency.

Rancangan Jaringan Highly Available PT Pundi Mas Berjaya (PMB)

In the digital age, networks are an important factor for implementing communication and exchanging information. Campus networks, which are private networks specifically designed for a particular institution, are essential for the institution because they provide access to the necessary information and services. However, campus networks are also vulnerable to security threats such as cybercrime and malware attacks. Firewall implementation can minimize these threats by controlling access to network services. Pundi Mas Berjaya (PMB), a company that provides software solutions to the global market, requires a highly available and redundant campus network. This research uses Cisco Packet Tracer to design and configure the network required by PMB. Implementing a highly available and redundant campus network with a hierarchical network model will improve the performance of the PMB campus network connectivity and security.

DRnet: Dynamic Retraining for Malicious Traffic Small-Sample Incremental Learning

Deep learning has achieved good classification results in the field of traffic classification in recent years due to its good feature representation ability. However, the existing traffic classification technology cannot meet the requirements for the incremental learning of tasks in online scenarios. In addition, due to the high concealment and fast update speed of malicious traffic, the number of labeled samples that can be captured is scarce, and small samples cannot drive neural network training, resulting in poor performance of the classification model. Therefore, this paper proposes an incremental learning method for small-sample malicious traffic classification. The method uses the pruning strategy to find the redundant network structure and dynamically allocates redundant neurons for training based on the proposed measurement method according to the difficulty of the new class. This enables the network to perform incremental learning without excessively consuming storage and computing resources, and reasonable allocation improves the classification accuracy of new classes. At the same time, through the knowledge transfer method, the model can reduce the catastrophic forgetting of the old class, relieve the pressure of training large parameters with small-sample data, and improve the model classification performance. Experiments involving multiple datasets and settings show that our method is superior to the established baseline in terms of classification accuracy, consuming 50% less memory.

Exploring the technological position and role of vehicle navigation companies by using patent citation network

Autonomous driving is the future trend when it comes to the age of 5G at the end of 21st century, and vehicle navigation is the indispensable key technology. The companies with car navigation software and hardware business are facing fierce competition. Business operators must know the company's position and role in the industry, in order to adopt strategies to respond. In this study, two indicators are investigated using the social network analysis technics such as technological redundancy and network centrality in patent citation network (PCN). It is understood that technological redundancy is determined by technology knowledge status (TKS) as well as technology knowledge reliability (TKR). There are also four types of social network centrality: degree centrality, eigenvector centrality, closeness centrality, and betweenness centrality. In this study, an analysis of the technical position and role of individual companies in the industry will be presented by using the two indicators and the primary components. Moreover, technological redundancy is used to study the technical position and role of a company's unitary patent capability in the industry. This will reinforce the company's understanding, and then take strategies. By using this model, the study can effectively identify the technical position and role of the company and its unitary patent capability. Furthermore, the study also shows that based on this model, there is a total of three companies where positions and roles are as leading status in the vehicle navigation industry, i.e. Tomtom map company, Alpine and Asin AW two electronic companies. In addition, companies with unique technologies, such as Sirius Xm Radio, etc. can use their role and position to conduct merger and acquisition, collaboration, and alliance strategies.

Revealing proteome-level functional redundancy in the human gut microbiome using ultra-deep metaproteomics

Functional redundancy is a key ecosystem property representing the fact that different taxa contribute to an ecosystem in similar ways through the expression of redundant functions. The redundancy of potential functions (or genome-level functional redundancy {{{{{{rm{FR}}}}}}}_{g}) of human microbiomes has been recently quantified using metagenomics data. Yet, the redundancy of expressed functions in the human microbiome has never been quantitatively explored. Here, we present an approach to quantify the proteome-level functional redundancy {{{{{{rm{FR}}}}}}}_{p} in the human gut microbiome using metaproteomics. Ultra-deep metaproteomics reveals high proteome-level functional redundancy and high nestedness in the human gut proteomic content networks (i.e., the bipartite graphs connecting taxa to functions). We find that the nested topology of proteomic content networks and relatively small functional distances between proteomes of certain pairs of taxa together contribute to high {{{{{{rm{FR}}}}}}}_{p} in the human gut microbiome. As a metric comprehensively incorporating the factors of presence/absence of each function, protein abundances of each function and biomass of each taxon, {{{{{{rm{FR}}}}}}}_{p} outcompetes diversity indices in detecting significant microbiome responses to environmental factors, including individuality, biogeography, xenobiotics, and disease. We show that gut inflammation and exposure to specific xenobiotics can significantly diminish the {{{{{{rm{FR}}}}}}}_{p} with no significant change in taxonomic diversity.

Predictive edge caching through deep mining of sequential patterns in user content retrievals

Edge caching plays an increasingly important role in boosting user content retrieval performance while reducing redundant network traffic. The effectiveness of caching ultimately hinges on the accuracy of predicting content popularity in the near future. However, at the network edge, content popularity can be extremely dynamic due to diverse user content retrieval behaviors and the low-degree of user multiplexing. It is challenging for the traditional reactive caching systems to keep up with the dynamic content popularity patterns. In this paper, we propose a novel Predictive Edge Caching (PEC) system that predicts the future content popularity using fine-grained learning models that mine sequential patterns in user content retrieval behaviors, and opportunistically prefetches contents predicted to be popular in the near future using idle network bandwidth. Through extensive experiments driven by real content retrieval traces, we demonstrate that PEC can adapt to highly dynamic content popularity at network edge, and significantly improve cache hit ratio and reduce user content retrieval latency over the state-of-art caching policies. More broadly, our study demonstrates that edge caching performance can be boosted by deep mining of user content retrieval behaviors.

Multi-Modality Adaptive Feature Fusion Graph Convolutional Network for Skeleton-Based Action Recognition

Graph convolutional networks are widely used in skeleton-based action recognition because of their good fitting ability to non-Euclidean data. While conventional multi-scale temporal convolution uses several fixed-size convolution kernels or dilation rates at each layer of the network, we argue that different layers and datasets require different receptive fields. We use multi-scale adaptive convolution kernels and dilation rates to optimize traditional multi-scale temporal convolution with a simple and effective self attention mechanism, allowing different network layers to adaptively select convolution kernels of different sizes and dilation rates instead of being fixed and unchanged. Besides, the effective receptive field of the simple residual connection is not large, and there is a great deal of redundancy in the deep residual network, which will lead to the loss of context when aggregating spatio-temporal information. This article introduces a feature fusion mechanism that replaces the residual connection between initial features and temporal module outputs, effectively solving the problems of context aggregation and initial feature fusion. We propose a multi-modality adaptive feature fusion framework (MMAFF) to simultaneously increase the receptive field in both spatial and temporal dimensions. Concretely, we input the features extracted by the spatial module into the adaptive temporal fusion module to simultaneously extract multi-scale skeleton features in both spatial and temporal parts. In addition, based on the current multi-stream approach, we use the limb stream to uniformly process correlated data from multiple modalities. Extensive experiments show that our model obtains competitive results with state-of-the-art methods on the NTU-RGB+D 60 and NTU-RGB+D 120 datasets.

Evaluating the Hybrid Multi-Protocol Label Switching (MPLS) on the Enhanced Interior Gateway Routing Protocol (EIGRP)

The development of technology and communication is expanding rapidly. In this case, the internet has become a vital necessity in globalization. Technological innovations are required to create a seamless, fast, and secure communication system. This research aims to evaluate the Enhanced Interior Gateway Routing Protocol (EIGRP) implementation by applying the Multiprotocol Label Switching (MPLS) technology. This study adhered to several stages in the Network Development Life Cycle (NDLC) method. The results of the two technology combinations, EIGRP and MPLS, demonstrated MPLS network simulation testing in several dynamic routing systems: EIGRP and OSPF, identified through the Quality of Service (QoS) value. It revealed that the best performance was EIGRP with a throughput of 2152.5 bps, delay of 335.6 ms, and jitter of 411 ms. Furthermore, MPLS and EIGRP network redundancy was better applied in the mesh topology with a multi or backup link than in the linear topology with a single link.

Distillation Sparsity Training Algorithm for Accelerating Convolutional Neural Networks in Embedded Systems

The rapid development of neural networks has come at the cost of increased computational complexity. Neural networks are both computationally intensive and memory intensive; as such, the minimal energy and computing power of satellites pose a challenge for automatic target recognition (ATR). Knowledge distillation (KD) can distill knowledge from a cumbersome teacher network to a lightweight student network, transferring the essential information learned by the teacher network. Thus, the concept of KD can be used to improve the accuracy of student networks. Even when learning from a teacher network, there is still redundancy in the student network. Traditional networks fix the structure before training, such that training does not improve the situation. This paper proposes a distillation sparsity training (DST) algorithm based on KD and network pruning to address the above limitations. We first improve the accuracy of the student network through KD, and then through network pruning, allowing the student network to learn which connections are essential. DST allows the teacher network to teach the pruned student network directly. The proposed algorithm was tested on the CIFAR-100, MSTAR, and FUSAR-Ship data sets, with a 50% sparsity setting. First, a new loss function for the teacher-pruned student was proposed, and the pruned student network showed a performance close to that of the teacher network. Second, a new sparsity model (uniformity half-pruning UHP) was designed to solve the problem that unstructured pruning does not facilitate the implementation of general-purpose hardware acceleration and storage. Compared with traditional unstructured pruning, UHP can double the speed of neural networks.