In this paper, a cross-spring flexure hinge is designed. The hinge comprises two leaf springs arranged in a cross configuration, with movement achieved through elastic deformation of the springs. By analysing the series-parallel relationship of the individual segments, an equivalent rotational stiffness model was established based on the elastic beam theory. Finite element simulations were conducted to evaluate the rotational stiffness, revealing that among the four design parameters, leaf thickness t, intersection hole edges a and b, and intersection angle α, the leaf thickness t has the most significant influence. Furthermore, three types of flexure hinge samples were manufactured using selective laser melting with AlSi10Mg, TC4, and 316L materials. Experimental results indicated that 316L provided the closest match to the simulated stiffness, with an average error of 2.34%. However, its high density limits its suitability for lightweight applications. TC4 demonstrates the best overall performance, balancing mechanical performance, print quality, and lightweight requirements. This research provides a valuable reference for the engineering design, material selection, and manufacturing process optimization of flexure hinges.

Accurately predicting microRNA-disease associations (MDAs) is crucial for identifying biomarkers and therapeutic targets in complex diseases. However, experimental validation is costly, and existing computational methods, particularly Graph Neural Networks (GNNs), are limited by data sparsity. In such sparse graphs, the lack of topological connections hinders effective message passing, leading to poor performance for isolated nodes and in cold-start scenarios. To address these challenges, we propose the adaptive dynamic heterogeneous graph attention neural network (ADHGMDA), a framework for MDAs prediction and analysis. First, to resolve connectivity issues, we construct a dual-layer heterogeneous graph enhanced with virtual nodes derived via K-means clustering. These virtual nodes act as semantic bridges, integrating isolated microRNAs and diseases into the network. Second, we introduce a dynamic feature learning mechanism that simulates data sparsity during training, forcing the model to learn robust representations for cold-start prediction. Third, to mitigate the over-smoothing common in deep GNNs, we design a gradient conflict-aware multi-task optimization strategy with dynamic weight adaptation. Furthermore, addressing the issue that existing benchmarks rely on outdated data, we reconstructed a benchmark dataset based on the latest databases, integrated with visualization tools. Experimental results demonstrate that ADHGMDA significantly outperforms seven state-of-the-art methods, achieving area under the receiver operating characteristic curve scores of 0.9698 and 0.9730. In case studies on herpes simplex and interstitial nephritis, the model exhibits excellent performance in uncovering potential pathogenic pathways. Meanwhile, extensive validation experiments confirm that it has good robustness against label noise, class imbalance, and the cold-start problem of isolated nodes.

Understanding the interactions between long non-coding RNAs (lncRNAs) and proteins is crucial for elucidating complex regulatory mechanisms within cells. Recent advances in graph-neural-network prediction methods have been significant. However, they still insufficiently address the sparsity of data in predicting associations between lncRNAs and proteins, and often overlook the importance of weighting multilevel representations during the encoding process of lncRNAs and proteins. Moreover, the high complexity of encoding networks can lead to model instability. We introduce a novel approach called the Graph residual connection attention Holistic processing multiple Channels graph Transformer (GHCT) for predicting interactions between lncRNAs and proteins. This model leverages identity feature matrices for adaptive encoding and employs deep graph residual modules to resolve issues of model instability. The GHCT captures complex feature representations and long-distance dependencies through holistic attentional processing. It also utilises this graph multichannel attention mechanism to capture cross-channel graph aggregation information and effectively capture and integrate information from different data channels. Finally, matrix multiplication is used for decoding and prediction. We conducted extensive experiments on three widely used datasets, including sparse datasets, in which our GHCT model achieved area under the curve (AUC) scores of 0.9775, 0.9746, and 0.9808, significantly outperforming other state-of-the-art models. Moreover, we conducted ten repeated experiments on imbalanced and noisy datasets, the results of which demonstrated the excellent generalisation and noise resistance capabilities of our model. Our case studies have also shown that our method can discover new lncRNA-protein interactions, providing valuable insights for predicting the interactions between lncRNAs and proteins.

Chatter has always been a key problem restricting the improvement of robotic milling quality and efficiency. To avoid chatter, it is necessary to determine what is the dominant chatter mechanism (mode coupling or regenerative) of the robot milling system. Therefore, this paper focus on the dominant chatter mechanism in high-load (600kg) robot milling. The modal test results show that the dynamic flexibility of spindle-tool structure mode in high-load robot is significantly higher than that of the body structure mode, which is significantly different from the low-load robot in other studies. The mode coupling chatter stability prediction models are established based on eigenvalue method and zeroth order approximation, and the predicted stability boundaries are compared with the experimental results. The results show that only high-frequency chatter exists in the high speed region (1000-8000rpm), and no low frequency chatter occurs. The low-frequency chatter around the robot body mode is found in the low-speed region (400-1000rpm), but the mode coupling chatter theory could not explain the chatter varies periodically with the spindle speed. However, the stability boundary predicted by the regenerative chatter theory also changes periodically with the spindle speed. This indicates that the milling chatter dominant mechanism of high load robot is regenerative chatter. This study analyzes the milling chatter dominant mechanism of high-load robot through theoretical and experimental verification, which can provide theoretical support for high-load robot milling chatter control.

With 6G's arrival, Inclusive Intelligent Services (IIS) require low-latency and high-reliability provisioning. This paper proposes GenAI-SFC, a Deep Reinforcement Learning (DRL) framework enhanced by a Conditional Variational AutoEncoder (CVAE) for adaptive Service Function Chaining (SFC). The CVAE improves generalization by modeling node state transitions, while the DRL agent optimizes Virtual Network Function (VNF) placement through trial-and-error learning. Simulation results show GenAI-SFC outperforms baseline algorithms, achieving lower provision cost, thereby enhancing performance for diverse and dynamic 6G application scenarios.

Passive binocular measurement systems are being increasingly utilized in the in-situ industries of automobiles, aviation, and aerospace, etc. due to their excellent qualities of accuracy, efficiency, and cost performance. Whereas the barrier of evaluating the accuracy of measured objects resulted from the unequal equivalent focal length and quantization of pixels, has limited their further development and application of high requirements for in-situ machining, e.g., the measurement of machining reference points for the positioning of robotic drilling in aerospace manufacturing. In this paper, an accuracy evaluation method is proposed to address the problem. Firstly, the unequal equivalent focal length is considered to improve the accuracy of 3D reconstruction. Next, the credibility probability model is developed to calculate the probability of the observed error in the public view of the binocular measurement system and indicates the direction of improvement. Finally, the in-situ experiment is carried out to validate the method within the effective public view range of 300 mm × 300 mm. The experiment results show that the RMSs of observed errors are superior to 0.035 mm, and the credibility probabilities are all higher than 0.91; the maximum 3D reconstruction accuracy improvement is 60.3%, with the error reduced from 0.078 mm to 0.031 mm.

The change of tool tip frequency response caused by the posture dependence of robot dynamic is one of the key problems that make it difficult to accurately predict the milling stability of robot. In this paper, a tool tip frequency response prediction method considering the interface stiffness characteristics of spindle-tool system is proposed for stability prediction of robotic milling under the condition of posture variation. Firstly, the interface stiffness models of spindle-toolholder, toolholder-spring clip and spring clip-tool are established based on Yoshimura's unit area method. Then, The dynamics model for the robot body and the interface stiffness models for spindle-tool system are imported into the finite element analysis model of the spindle system, so that the prediction of the tool tip frequency response is realized by harmonic response analysis. Compared with the experimental results, the maximum error of the natural frequency was not more than 2 %, and the maximum error of the amplitude was not more than 12%. Finally, the 2 DOF robot milling stability prediction model is established. Then the robot milling chatter is predicted considering redundant degrees of freedom from the perspective of regenerative chatter prediction theory, and the accuracy of prediction results is verified by milling experiment.

The Atribacterota are widely distributed in the subsurface biosphere. Recently, the first Atribacterota isolate was described and the number of Atribacterota genome sequences retrieved from environmental samples has increased significantly; however, their diversity, physiology, ecology, and evolution remain poorly understood. We report the isolation of the second member of Atribacterota, Thermatribacter velox gen. nov., sp. nov., within a new family Thermatribacteraceae fam. nov., and the short-term laboratory cultivation of a member of the JS1 lineage, Phoenicimicrobium oleiphilum HX-OS.bin.34TS, both from a terrestrial oil reservoir. Physiological and metatranscriptomics analyses showed that Thermatribacter velox B11T and Phoenicimicrobium oleiphilum HX-OS.bin.34TS ferment sugars and n-alkanes, respectively, producing H2, CO2, and acetate as common products. Comparative genomics showed that all members of the Atribacterota lack a complete Wood-Ljungdahl Pathway (WLP), but that the Reductive Glycine Pathway (RGP) is widespread, indicating that the RGP, rather than WLP, is a central hub in Atribacterota metabolism. Ancestral character state reconstructions and phylogenetic analyses showed that key genes encoding the RGP (fdhA, fhs, folD, glyA, gcvT, gcvPAB, pdhD) and other central functions were gained independently in the two classes, Atribacteria (OP9) and Phoenicimicrobiia (JS1), after which they were inherited vertically; these genes included fumarate-adding enzymes (faeA; Phoenicimicrobiia only), the CODH/ACS complex (acsABCDE), and diverse hydrogenases (NiFe group 3b, 4b and FeFe group A3, C). Finally, we present genome-resolved community metabolic models showing the central roles of Atribacteria (OP9) and Phoenicimicrobiia (JS1) in acetate- and hydrocarbon-rich environments. Our findings expand the knowledge of the diversity, physiology, ecology, and evolution of the phylum Atribacterota. This study is a starting point for promoting more incisive studies of their syntrophic biology and may guide the rational design of strategies to cultivate them in the laboratory.

Through the integration of Network Function Virtualization (NFV), Software-Defined Networks, cloud service providers can realize cost reduction, performance enhancement, and support for advanced application scenarios. Within this framework, Service Function Chain (SFC) has emerged as a pivotal approach for delivering network services, offering flexibility and cost-effectiveness in provisioning Virtual Network Functions (VNFs) dynamically and elastically across a network of physical devices. This paper concentrates on optimizing VNF placement within NFV-enabled networks to elevate the number of SFC Requests (SFCRs) successfully processed. Emphasizing the often-overlooked factors of time-varying workloads and VNF sharing via multi-tenancy technology, we present a novel Integer Linear Programming (ILP) formulation for the strategic placement of VNFs. To effectively solve this ILP, we introduce a Throughput Optimization Heuristic Solution (TOHS), comprising an algorithm based on correlation for assigning SFCRs to nodes, alongside a VNFR aggregation algorithm aimed at optimizing node resource utilization by consolidating VNFRs requiring identical VNF types. Through detailed numerical simulations and comparative analysis with existing literature, we demonstrate that TOHS approaches the performance of ILP solvers for small-scale problems, outperforming other solutions in diverse scenarios. Specifically, TOHS achieves higher network throughput and more efficient utilization of network resources, underscoring its efficacy in practical NFV-enabled network deployments.

The global expansion of 5G networks has led to a significant increase in network traffic. In data centers, virtual machines (VMs) must be allocated on Physical Machines (PMs) according to a specific topology. Each VM requires specific network resources to function correctly. Consolidated VM deployment can help reduce traffic consumption and prevent bandwidth-related bottlenecks, while loose deployment can minimize VM failure rates and guarantee availability during PM and switch failures. A reasonable VM deployment plan is vital to improve availability and minimize network bandwidth consumption. This paper presents four typical data center architectures, network topologies, and cost matrices extending to generality. A joint optimization model is proposed to measure Virtual Cluster (VC) risk with global availability constraints. A heuristic algorithm is then introduced to minimize the value of the constrained optimization function. The evaluation results indicate that the proposed method is effective and improves performance over the benchmarks.

The advancement of artificial intelligence (AI) has the potential to revolutionize network communication. The use of advanced feature extraction in semantic communication can enhance transmission capacity. However, relying solely on unimodal visual characteristics derived from images through these approaches may result in inaccuracies in decoding under low signal-to-noise ratio (SNR) conditions. This paper introduces LaMoSC, a semantic communication system driven by large language models (LLMs) that uses multimodal features to reconstruct raw visual information, thereby improving transmission quality. The system proposes an LLM-driven multimodal fusion semantic communication framework, which aims to expand unimodal transmission systems and enhance generalization ability. LaMoSC has designed an end-to-end encoding-decoding network that integrates visual and textual multimodal feature inputs. The design deeply integrates modal features using the attention mechanism. Comprehensive comparisons with state-of-the-art baselines across various datasets demonstrate the robustness of the proposed method, particularly highlighting its superiority in low SNR conditions. LaMoSC outperforms Deep-JSCC and multi-level semantic aware communication system (MLSC) by 5.5% and 2.6%, respectively, under low SNR conditions, such as 4 dB. Its exceptional generalization capacity sets it apart from other methods.

Background In cellular activities, essential proteins play a vital role and are instrumental in comprehending fundamental biological necessities and identifying pathogenic genes. Current deep learning approaches for predicting essential proteins underutilize the potential of gene expression data and are inadequate for the exploration of dynamic networks with limited evaluation across diverse species. Results We introduce ECDEP, an essential protein identification model based on evolutionary community discovery. ECDEP integrates temporal gene expression data with a protein–protein interaction (PPI) network and employs the 3-Sigma rule to eliminate outliers at each time point, constructing a dynamic network. Next, we utilize edge birth and death information to establish an interaction streaming source to feed into the evolutionary community discovery algorithm and then identify overlapping communities during the evolution of the dynamic network. SVM recursive feature elimination (RFE) is applied to extract the most informative communities, which are combined with subcellular localization data for classification predictions. We assess the performance of ECDEP by comparing it against ten centrality methods, four shallow machine learning methods with RFE, and two deep learning methods that incorporate multiple biological data sources on Saccharomyces. Cerevisiae (S. cerevisiae) , Homo sapiens (H. sapiens) , Mus musculus , and Caenorhabditis elegans . ECDEP achieves an AP value of 0.86 on the H. sapiens dataset and the contribution ratio of community features in classification reaches 0.54 on the S. cerevisiae (Krogan) dataset. Conclusions Our proposed method adeptly integrates network dynamics and yields outstanding results across various datasets. Furthermore, the incorporation of evolutionary community discovery algorithms amplifies the capacity of gene expression data in classification.

Low-latency, real-time rendering of 3D objects is critical for mobile web-based augmented reality (MWAR) applications. While cloud-based or edge-based server rendering offloading can reduce the latency for mobile devices, a high volume of user requests in user aggregation scenarios can overload servers and transmission channels. It can result in a poor user experience due to resource consumption and high latency. This paper presents a decentralized and collaborative real-time rendering offloading network architecture (CRCDnet) to address this issue. CRCDnet makes contributions in the following three areas: (1) In the user aggregation scenarios, mobile devices that perform the same services are used as service nodes to perform the offloading of rendering computing and a collaborative rendering computing network is established. (2) For data exchange in decentralized networks, a data sharing middle layer based on blockchain key indexes separates sensitive service data and ensures a secure, reliable exchange mechanism and efficient data exchange. (3) A data request and computing offloading scheduling approach is proposed for the collaborative rendering computing network to optimize the rendering computation delay.

In the 6G vision, networks are expected to be more flexible in quickly solving network traffic scheduling issues and deploying services. Network Function Virtualization (NFV) is an innovative technology that involves extracting network functions from dedicated equipment to create Virtual Network Functions (VNFs). These VNFs are then chained together to form a Service Function Chain (SFC) that provides network service. However, there are still some issues with existing network service orchestration tools, such as unreasonable multi-traffic scheduling and additional programming requirements for end-users. We have developed a solution to address the challenges posed by data coupling and bandwidth preemption in multi-service environments. Our dynamic Quality of Service (QoS) Guarantee model utilizes hierarchical analysis to prioritize traffic among multiple service data streams and employs a service scheduling algorithm based on a weighted fair queue to allocate link resources. For user convenience, we have also created an intuitive web orchestration platform called EasyOrchestrator, enabling users to encapsulate common VNFs and build services quickly. Our experimental evaluation has shown that EasyOrchestrator significantly reduces service construction time compared to the benchmark. At the same time, our QoS assurance mechanism effectively minimizes network congestion and ensures the successful operation of high-priority services.

The emergence of Software-Defined Networks (SDN) and Network Function Virtualization (NFV) has made Service Function Chain (SFC) a popular method for delivering network services. This innovative computing and networking paradigm allows Virtual Network Functions (VNFs) to be cost-effectively deployed on a network of physical equipment flexibly and elastically. Traffic can be directed as needed by linking VNFs as an SFC. However, the current algorithms for VNF placement computation and traffic steering in SFC are often complex, unscalable, and time-consuming. This paper investigates the VNF placement and SFC chaining problem in NFV-enabled networks. To obtain the VNF placement solution that maximizes network resource utilization, we formulate the problem as a Binary Integer Programming (BIP) model. Additionally, we introduce a novel Deep Learning-based VNF Placement Algorithm (DLVPA) that uses an intelligent node selection network to place VNFs for SFC requests. Performance evaluations demonstrate that DLVPA can effectively improve network resource utilization and achieve high solution computation time efficiency.

BACKGROUND:The association between MicroRNAs (miRNAs) and diseases is crucial in treating and exploring many diseases or cancers. Although wet-lab methods for predicting miRNA-disease associations (MDAs) are effective, they are often expensive and time-consuming. Significant advancements have been made using Graph Neural Network-based methods (GNN-MDAs) to address these challenges. However, these methods still face limitations, such as not considering nodes' deep-level similarity associations and hierarchical learning patterns. Additionally, current models do not retain the memory of previously learned heterogeneous historical information about miRNAs or diseases, only focusing on parameter learning without the capability to remember heterogeneous associations. RESULTS:This study introduces the K-means disentangled high-level biological similarity to utilize potential hierarchical relationships fully and proposes a Graph Attention Heterogeneous Biological Memory Network architecture (DiGAMN) with memory capabilities. Extensive experiments were conducted across four datasets, comparing the DiGAMN model and its disentangling method against ten state-of-the-art non-disentangled methods and six traditional GNNs. DiGAMN excelled, achieving AUC scores of 96.35%, 96.10%, 96.01%, and 95.89% on the Data1 to Data4 datasets, respectively, surpassing all other models. These results confirm the superior performance of DiGAMN and its disentangling method. Additionally, various ablation studies were conducted to validate the contributions of different modules within the framework, and's encoding statuses and memory units of DiGAMN were visualized to explore the utility and functionality of its modules. Case studies confirmed the effectiveness of DiGAMN's predictions, identifying several new disease-associated miRNAs. CONCLUSIONS:DiGAMN introduces the use of a disentangled biological similarity approach for the first time and successfully constructs a Disentangled Graph Attention Heterogeneous Biological Memory Network model. This network can learn disentangled representations of similarity information and effectively store the potential biological entanglement information of miRNAs and diseases. By integrating disentangled similarity information with a heterogeneous attention memory network, DiGAMN enhances the model's ability to capture and utilize complex underlying biological data, significantly outperforming many existing models. The concepts used in this method also provide new perspectives for predicting miRNAs associated with diseases.

Abstract Kiwifruit is an economically and nutritionally important fruit crop with extremely high contents of vitamin C. However, the previously released versions of kiwifruit genomes all have a mass of unanchored or missing regions. Here, we report a highly continuous and completely gap-free reference genome of Actinidia chinensis cv. ‘Hongyang’, named Hongyang v4.0, which is the first to achieve two de novo haploid-resolved haplotypes, HY4P and HY4A. HY4P and HY4A have a total length of 606.1 and 599.6 Mb, respectively, with almost the entire telomeres and centromeres assembled in each haplotype. In comparison with Hongyang v3.0, the integrity and contiguity of Hongyang v4.0 is markedly improved by filling all unclosed gaps and correcting some misoriented regions, resulting in ~38.6–39.5 Mb extra sequences, which might affect 4263 and 4244 protein-coding genes in HY4P and HY4A, respectively. Furthermore, our gap-free genome assembly provides the first clue for inspecting the structure and function of centromeres. Globally, centromeric regions are characterized by higher-order repeats that mainly consist of a 153-bp conserved centromere-specific monomer (Ach-CEN153) with different copy numbers among chromosomes. Functional enrichment analysis of the genes located within centromeric regions demonstrates that chromosome centromeres may not only play physical roles for linking a pair of sister chromatids, but also have genetic features for participation in the regulation of cell division. The availability of the telomere-to-telomere and gap-free Hongyang v4.0 reference genome lays a solid foundation not only for illustrating genome structure and functional genomics studies but also for facilitating kiwifruit breeding and improvement.

Robot end-effector positioning is crucial for performing high-accuracy tasks. This article proposes a novel method for achieving online high-accuracy positioning of the spindle end based on the eye-in-hand active correction. The positioning process includes the working area locking based on the feature recognition and the coordinate high-accuracy positioning. For the latter one, first, a new hand–eye calibration approach is introduced to acquire the relationship between the cameras and the robot end without being constrained by the public field of view (FoV) of the binocular cameras to establish accurate transformation for positioning process. Second, the spindle end is positioned to achieve an accurate position by repeatedly updating the difference between the current position and the preset coordinate with the active correction of the eye-in-hand vision assisted by local machining reference points. Experimental results indicate that the proposed method can reduce the absolute positioning error of the spindle end from 0.098 to 0.078 mm, which is reduced by 20% compared with the laser tracker (LT) method.

Image datasets acquired from orchards are commonly characterized by intricate backgrounds and an imbalanced distribution of disease categories, resulting in suboptimal recognition outcomes when attempting to identify apple leaf diseases. In this regard, we propose a novel apple leaf disease recognition model, named RFCA ResNet, equipped with a dual attention mechanism and multi-scale feature extraction capacity, to more effectively tackle these issues. The dual attention mechanism incorporated into RFCA ResNet is a potent tool for mitigating the detrimental effects of complex backdrops on recognition outcomes. Additionally, by utilizing the class balance technique in conjunction with focal loss, the adverse effects of an unbalanced dataset on classification accuracy can be effectively minimized. The RFB module enables us to expand the receptive field and achieve multi-scale feature extraction, both of which are critical for the superior performance of RFCA ResNet. Experimental results demonstrate that RFCA ResNet significantly outperforms the standard CNN network model, exhibiting marked improvements of 89.61%, 56.66%, 72.76%, and 58.77% in terms of accuracy rate, precision rate, recall rate, and F1 score, respectively. It is better than other approaches, performs well in generalization, and has some theoretical relevance and practical value.