Accurate localization of human acupoints remains a fundamental challenge in the development of massage robot systems. While deep learning algorithms leveraging the classical Transformer architecture have improved recognition accuracy, a notable gap still exists between current performance and the robustness needed to handle complex lighting conditions and individual variability. To fill this gap, this article focuses on the scenarios involving low-light conditions and subjects wearing close-fitting clothing. To address this issue, this article proposes an acupoint localization algorithm tailored to the scenario by employing multimodal data fusion techniques that integrate thermal imaging with depth map information. The algorithm is based on YOLOv8 and has been architecturally improved, referred to as YOLOv8-TCC. Specifically, a dual-branch parallel backbone network architecture is designed, incorporating the channel prior convolutional attention (CPCA) module to enhance multimodal feature representation. Next, a cross transformer fusion (CTF) module is introduced to facilitate information integration, while a dynamic template alignment approach is proposed to accommodate variations in body shape and posture. Experimental results show that the YOLOv8-TCC method achieves higher accuracy compared to the traditional single-modal model under varying lighting conditions. With the introduction of the CTF and CPCA modules, the average precision (AP) at intersection over union (IoU) (AP@0.75) increased by 2.8% and 0.8%, respectively. Moreover, the success rate under the clinically meaningful 0.5 cun constraint (SR@0.5 cun) reached 95%. It is worth mentioning that our method remains effective when subjects wear light and close-fitting clothing. The proposed method exhibits strong generalizability and scalability, offering an accurate and efficient solution for practical applications.

To address the challenges in Industrial Internet of Things (IIoT) systems—stemming from heterogeneous device computing capabilities and diverse network topologies, which often result in excessive resource consumption and inefficiencies in traditional blockchain consensus algorithms—this paper proposes a Node-Efficient Consensus Algorithm for topology-heterogeneous large-scale IIoT Systems based on a Multichain Directed Acyclic Graph (NECA-MDAG). First, we design a blockchain architecture based on multichain directed acyclic graph (DAG) with a hierarchical structure comprising five layers: device, edge, network, consensus, and application. This design facilitates optimized data sharing between multiple factories or organizational departments. Next, taking into account heterogeneous network topologies—including tree, mesh, and hybrid structures—we propose a lightweight multichain DAG model based on regional division. The nodes construct local slave DAG chains alongside a master DAG chain, thus enhancing the overall efficiency of the consensus process. We then propose an efficient consensus algorithm that combines local slave chain consensus with master chain consensus, enabling parallel consensus across distributed regions and substantially improving transaction throughput. Additionally, we develop a bidirectional cross-chain anchoring mechanism to facilitate secure, cross-regional sharing of industrial data, including inventory and order information. Experimental results show that, irrespective of node scale or network latency, NECA-MDAG consistently enhances transaction throughput, reduces latency and communication overhead, and surpasses CDBFT, LNLCA, and Avalanche in overall performance within IIoT environments.

Trust management in decentralized environments is critical for mitigating risks in autonomous interactions. However, existing blockchain-based reputation systems often suffer from two major limitations: (1) scalar reductionism, where complex trust attributes are compressed into a single static score, failing to capture the multidimensional nature of trust; and (2) lack of context awareness, where the same reputation value is applied indiscriminately across heterogeneous scenarios (e.g., C2C vs. G2B). In this paper, we propose UniTrust, a novel on-chain trust management framework. UniTrust introduces the concept of a "Trust Hypercube," decoupling trust into four vector dimensions: Identity, Capability, Behavior, and Reputation. We further design a generic Context Adapter mechanism that dynamically adjusts trust weights based on interaction types. To ensure system integrity without sacrificing privacy, we implement a granular auditing mechanism using Merkle Hash Trees (MHT), enabling lightweight on-chain verification of trust updates. Simulation results demonstrate that UniTrust effectively mitigates Sybil attacks and achieves superior threshold-based discrimination between honest and malicious agents, maintaining a wider trust score gap across heterogeneous interaction contexts compared to traditional scalar reputation models.

We model cryptographic auditing of off-chain data as a Constrained MDP (CMDP) under partial observability: the storage node's hidden type and corruption state make the problem a POMDP, while a miss-rate ceiling rho imposes an explicit security constraint. We propose DRQN-CMDP, a Deep Recurrent Q-Network whose GRU layer maintains a belief over the latent node type, paired with Lagrangian dual ascent that adapts the miss-rate penalty lambda automatically. A pairing-free homomorphic-MAC primitive supplies O(1) on-chain verification cost. Across 13 methods–four DQN variants, PPO, A2C, PPO-Lagrangian, a stateful Bayesian heuristic, three fixed-rule baselines, and an oracle-informed heuristic–DRQN-CMDP achieves a favourable balance: 83

As blockchain technology extends into societal-scale applications, dynamic service interactions (e.g., e-commerce, e-government, and social computing) are becoming increasingly complex. This renders traditional single-dimensional trust metrics insufficient for the security demands of heterogeneous Web3 service ecosystems. This paper presents a multi-level node trust evaluation framework named Context-Aware Trust Graph Algebra (CTAG). Based on algebraic graph theory, the framework defines a formal trust graph model and characterizes dynamic interaction environments precisely through the introduction of a "Scene Sextuplet." The core contribution of CTAG lies in overcoming the limitations of traditional context-agnostic models, seamlessly supporting diverse on-chain service interaction scenarios, including C2C, B2C, B2B, G2B, and G2C. By employing three distinct operators—Trust Decay, Propagation, and Composition—CTAG addresses the issue of semantic loss during cross-scene trust transmission. Furthermore, to address the challenge of trust security level classification, the model integrates Privacy Level and Governance Rule parameters. This enables fine-grained authorization management, ensuring the dynamic adjustment of trust policies across contexts with varying security requirements. We implemented a high-performance system component (TrustEngine) in Python, which encapsulates graph operation interfaces for context-aware trust evaluation. Experimental results on controlled deterministic scenarios show that CTAG exhibits context-sensitive behaviors, including cross-scene attenuation, multi-hop propagation, multi-path composition, and stable scene extension, while maintaining low query latency compared with a topology-based PageRank baseline in the tested setting. Dynamic scene extension tests further indicate that CTAG can preserve existing scene-specific trust values when new service contexts are introduced. This study provides an interpretable algebraic basis and core algorithmic components for scalable distributed identity authentication and service trust management systems.

In this paper, a Multiple-Objective Memetic Algorithm (MOMA) is proposed to address the Energy-Efficient Distributed Assembly Permutation Flow-Shop Scheduling Problem (EEDAPFSP) by explicitly exploiting the structural and objective symmetries inherent in the scheduling process, with the dual objectives of minimizing the maximum completion time (makespan) and total energy consumption (TEC). The EEDAPFSP is a complex NP-hard optimization problem in modern sustainable manufacturing that balances production efficiency and environmental sustainability. During the global search phase, a symmetry-preserving dual-search framework is constructed, in which diverse and potential regions in the solution space are explored by symmetrically generating time-dominant product sub-sequences (TDPSs) and energy-dominant product sub-sequences (EDPSs) in the individuals of each iteration, enabling complementary exploration from time- and energy-oriented perspectives. This is accomplished through the incorporation of a variable-weight metric technique and a first product fixed strategy into an estimation distributed algorithm-based hyper-heuristic (EDAHH), so as to maintain a balanced and symmetric probabilistic modeling of decision patterns with respect to the makespan and energy consumption. In the local search phase, two problem-specific designed neighborhood structures are proposed to refine the job sequences corresponding to the TDPS and EDPS in the superior sub-population, effectively reducing both the makespan and TEC. A box-level epsilon dominance technique based on the crowding distance is proposed for Pareto archive updating. Additionally, an energy-saving strategy is embedded throughout the algorithm, incorporating three mechanisms-job processing delay, machine shutdown and restart control, and speed regulation-to further optimize TEC during both the global and local search phases. Finally, extensive computational experiments are carried out, and the results demonstrate that the MOMA achieves significantly better performance in terms of the inverted generational distance (IGD) and the quality metric rho compared with state-of-the-art algorithms. The resulting Pareto front of non-dominated solutions provides a comprehensive set of trade-offs between energy consumption and the makespan, offering decision makers flexible and efficient scheduling options.

Blockchain oracle aggregation must fuse independently authenticated off-chain reports into a trusted estimate, yet existing methods address only one security dimension: signature schemes cannot detect bias injected by legitimately registered keys, while Byzantine-fault-tolerant aggregators assume static, non-forgeable identities. This joint gap motivates a unified admission pipeline.We propose a layered framework integrating hard Ed25519 authentication gates (signature verification and nonce/timestamp freshness) with a Beta-inspired Bayesian reputation accumulator and MAD-scaled consistency scorer, regulated by a three-state admission controller (normal/probation/exclusion). Gate 1 authenticates signed report provenance; it does not prove off-chain data value correctness.Experiments across 10 seeds show the framework matches Krum in static fixed-offset settings (Δ=0.0018, p=0.76, ns) while significantly outperforming all baselines under dynamic threats— behavioral change, intermittent attacks, and heavy-tail noise (p<0.001, Cohen’s d≥2.59). Disabling signature authentication degrades MAE by 8.28×, confirming its role as an indispensable prerequisite for behavioral trust. We further derive and validate the closed-form detectability boundary ${\delta ^{\ast}} = - {\hat s_{{\text{rob }}}}\ln {\tau _R} \approx 3.57$; strikingly, cumulative attack damage peaks precisely at δ*, revealing this point as simultaneously the system’s detection limit and the adversary’s damage-optimal bias—an emergent property of the joint authentication– reputation design.

To address the prevalent challenges in the Wireless Internet of Things (WIoT)—namely, the severe resource constraints of massive heterogeneous nodes, time-varying network environments, exorbitant overhead, low throughput, and storage bottlenecks inherent in traditional blockchain consensus mechanisms— we propose a novel block consensus and on-chain pruning algorithm for WIoT nodes, denoted LEAP-DAG. Initially, a data preprocessing mechanism based on hybrid verification and conflict arbitration is proposed. By calculating temporal and spatial dispersion degrees for multi-dimensional cross-validation and integrating causality-based voting arbitration within subnets, this mechanism effectively eliminates logically conflicting data. Subsequently, a dynamic neighbor committee election strategy is designed based on comprehensive scoring. This strategy evaluates the Round-Trip Time (RTT), packet loss rate, and computational response latency of nodes to dynamically filter high-quality nodes for the constitution of a consensus committee, which employs pipelined parallel verification to achieve consensus. Finally, a hierarchical Directed Acyclic Graph (DAG) transaction block on-chain integration and pruning mechanism is proposed. By categorizing transaction priorities to architect a hierarchical topology, this mechanism executes secure pruning of obsolete routine data on an epoch basis. This resolves the storage overflow predicament for ordinary gateway nodes while preserving complete historical traceability through full-ledger gateway nodes. The experimental results clearly demonstrate that LEAP-DAG consistently outperforms IOTA, SG-PBFT, and BCT in terms of transaction throughput, transaction confirmation latency, communication overhead, and storage saving ratio. Furthermore, it exhibits extraordinary robustness under large-scale malicious node attack scenarios.

This paper studies the Distributed Assembly Permutation Flowshop Scheduling Problem with Sequence Dependent Setup Times (DAPFSP-SDST). The optimization objective is minimization of maximal completion time (makespan), and it is shown that minimizing the sum of Total Setup Times for Assembling Products (TSTAP) and Total Idle Times on Assembly Machine (TITAM) is equivalent to minimizing the makespan. Additionally, minimization of TSTAP and TITAM can be transformed into sequencing problems of products and jobs within critical products, respectively. We also find out that a product sequence essentially explores an area in the solution space, with solutions in the area having the same TSTAP but different TITAMs determined by Critical-Jobs-Sequences (CJSs). Based on the new findings, an effective Two-Stage Heuristic Algorithm (TSHA) is proposed to first obtain promising product sequences and then exploit the corresponding areas for the DAPFSP-SDST. At the first stage of TSHA, high-quality initial product sequences are obtained through a constructive method and two Neighborhood Descent for Product Sequence (NDPS) algorithms are presented for more potential searching areas. At the second stage, a Neighborhood Descent for CJS (NDCJS) is designed to find CJSs with TITAM as small as possible. Evaluations on a benchmark instance set show that TSHA achieves better performance compared to the existing meta-heuristic and hyper-heuristic algorithms on the DAPFSP-SDST. Another impressive advantage of the TSHA is its low computational cost, as it is a heuristic algorithm with some neighborhood search operators

Graph representation learning has emerged as a powerful approach for modeling structured data across diverse domains, including social networks, biochemical interactions, and financial transaction systems. Existing contrastive learning frameworks have demonstrated strong performance in self-supervised graph learning; however, they often suffer from suboptimal augmentations, inadequate structural preservation, and a lack of adaptability to heterogeneous graph structures such as blockchain transaction networks. In this work, we propose ACLGMI (Adaptive Contrastive Learning with Graph Mutual Information Maximization), a novel framework that extends traditional contrastive learning by introducing an adaptive augmentation strategy and a mutual information maximization objective to enhance representation quality. ACLGMI dynamically adjusts augmentation strategies based on graph topology, ensuring that critical structural properties are preserved while improving robustness to noise and adversarial perturbations. Furthermore, we introduce a multi-level mutual information constraint that maximizes global and local consistency in learned representations, leading to improved performance across multiple downstream tasks. We evaluate ACLGMI on benchmark datasets for graph classification and on the Elliptic Bitcoin dataset for blockchain fraud detection. The results demonstrate that ACLGMI consistently outperforms state-of-the-art unsupervised methods, achieving superior accuracy in graph classification and higher ROC-AUC scores in fraud detection. Our findings suggest that ACLGMI provides a generalizable and robust solution for structured data learning, with strong applications in blockchain security and beyond.

Graph Convolutional Networks (GCNs) have achieved remarkable success in recommendation systems by leveraging higher-order neighborhoods. In recent years, multi-behavior recommendation has addressed the challenges of data sparsity and cold start problems to some extent. However, the introduction of noise from multi-behavior tasks into the user-item graph exacerbates the impact of noise from a few active users and popularity bias from popular items. To tackle these challenges, graph augmentation has emerged as a promising approach in recommendation systems. However, existing augmentation methods may generate suboptimal graph structures, and maximizing correspondence may capture information unrelated to the recommendation task. To address these issues, we propose a novel approach called the Multi-Behavior Adaptive Graph Contrastive Learning Model (MB-AGCL) for recommendation. Our approach integrates auxiliary behaviors to compensate for data sparsity and utilizes adaptive learning to determine whether to drop edges or nodes, thus obtaining an optimized graph structure that reduces the impact of noise. We then train the original and generated graphs using supervised tasks. Furthermore, we propose an efficient adaptive graph augmentation method that integrates graph augmentation with down-stream tasks to reduce the impact of popularity bias. Finally, we jointly optimize these two tasks. Through extensive experiments on public datasets, we validate the effectiveness of our recommendation model.

To address the problem of low positioning accuracy for long-distance static targets, we propose an optimized algorithm for long-distance target localization (LTLO) based on single-robot moving path planning. The algorithm divides the robot's movement area into hexagonal grids and introduces constraints on stopping position selection and non-redundant locations. Based on image parallelism, we propose a method for calculating the relative position of the target using sensing information from two positions. Additionally, an improved hierarchical density-Based spatial clustering of applications with noise (HDBSCAN) algorithm is developed to fuse the relative coordinates of multiple targets. Furthermore, we establish the corresponding constraints for long-distance target localization and construct a target localization optimization model based on single-robot path planning. To solve this model, we employ a double deep Q-network and propose a reward strategy based on coordinate fusion error. This approach solves the optimization model and obtains the optimal target positions and path trajectories, thereby improving the positioning accuracy for long-distance targets. Experimental results demonstrate that for static targets at distances ranging from 100 to 500 meters, LTLO outperforms traditional monocular visual localization (TMVL), monocular global geolocation (MGG) and long-range binocular vision target geolocation (LRBVTG) by obtaining an optimal path to identify target positions, maintaining a relative localization error within 4% and an absolute localization error within 6%.

To solve issues of frequent network topology changes to slow consensus speed and low security in consensus algorithms within heterogeneous dynamic Internet of Things (IoT) systems, we propose a fast consensus algorithm of large-scale heterogeneous dynamic IoT nodes for directed acyclic graph (DAG)-based blockchain (LSHD_DAG). First, an adaptive regional division method for heterogeneous dynamic nodes is proposed, which monitors nodes' state and counts within each region, adjusting the boundaries of adjacent regions. Second, an event-triggered main-sub DAG chain design is proposed. When transaction information pertains to a single region, the transaction is processed as a local transaction to construct a regional DAG sub chain. If transaction information involves multiple regions, it is organized into a multiregional transaction set and incorporated into the global DAG main chain. Next, a node grade identification mechanism that balances state and reputation factors is proposed. This mechanism adopts the level of participation (LOF) algorithm to classify or dynamically update node grades by considering node state and reputation evaluation results, assigning them identity permissions. Finally, a weighted voting consensus based on the main-sub DAG chain is introduced. Nearby honest nodes with stable movement are selected as consensus nodes, which perform multiregional weighted voting consensus on transactions and transaction sets. This process concludes with two-way pegging with multisignature on the main-sub DAG, ensuring secure and rapid consensus for data in heterogeneous and dynamic IoT environments. The experimental results show that no matter how the number of nodes changes, LSHD_DAG can improve transaction throughput and reduce latency and communication overhead, outperforming the credit-delegated byzantine fault tolerance (CDBFT), DAG double (DAG-D), and Avalanche.

In order to solve the problems in the blockchain consensus algorithm, such as slow consensus speed, high resource consumption, and low consensus security due to excessive large-scale node data volume and Byzantine attacks in the Internet of Things (IoT) system, we propose a large-scale node lightweight consensus algorithm of blockchain for IoT (LNLCA). First, a transaction set construction mechanism is proposed to improve consensus efficiency and security. The mechanism packages multiple data monitored by IoT nodes into data transactions, and builds the transaction set. Then, it constructs nonconflict subsets and conflict subsets. Second, an adjacent parent node sampling and response mechanism is proposed to reduce the consumption of communication resources. The new transaction set selects nearby parent nodes based on the time distance summation method and collaborates to batch verify each transaction in the transaction set. Finally, an efficient consistent consensus based on transaction set directed acyclic graph (DAG) is proposed to quickly vote for each transaction in the transaction set, thereby constructing the transaction set DAG for batch uploading to the chain, and achieving a secure and lightweight consensus on IoT data. The experimental results show that no matter how the number of Byzantine nodes changes, LNLCA can improve transaction throughput, and reduce transaction delay and communication overhead, which outperforms the credit-delegated Byzantine fault tolerance, Avalanche, and Hashgraph.

In this paper, a Hybrid Tabu Search (HTS) algorithm is proposed to solve the Distributed Assembly Permutation Flowshop Scheduling Problem with Sequence Dependent Setup Times (DAPFSP-SDST). For the objective of minimizing the maximal completion time (makespan), the optimization process is divided into two steps to minimize the Sum of Assembly Setup Times (SAST) and the Sum of Idle Time for Assembly Machine (SITAM), respectively. This is based on finding that minimization of makespan equals to minimization of the sum of SAST and SITAM for the DAPFSP-SDST. At the first step, a constructive algorithm is designed to obtain initial product sequences for the sake of SAST. Then, a tabu search algorithm is presented to perform on the product sequences for searching more promising areas in the solution space with both smaller SAST and various critical product sequences. At the second step, a critical-job-sequence-based local search algorithm is proposed to exploit these areas to find solutions with SITAM as small as possible. Finally, Experiment results show the effectiveness of the proposed HTS algorithm.

This paper introduces the efficient medical-images-aimed segment anything model (EMedSAM), addressing the high computational demands and limited adaptability of using SAM for medical image segmentation tasks. We present a novel, compact image encoder, DD-TinyViT, designed to enhance segmentation efficiency through an innovative parameter tuning method called med-adapter. The lightweight DD-TinyViT encoder is derived from the well-known ViT-H using a decoupled distillation approach.The segmentation and recognition capabilities of EMedSAM for specific structures are improved by med-adapter, which dynamically adjusts the model parameters specifically for medical imaging. We conducted extensive testing on EMedSAM using the public FLARE 2022 dataset and datasets from the First Hospital of Zhejiang University School of Medicine. The results demonstrate that our model outperforms existing state-of-the-art models in both multi-organ and lung segmentation tasks.

Addressing issues such as low efficiency in feature extraction, suboptimal feature quality, and low identification accuracy, this paper proposes an innovative Time Series Identification Method (TSIM) of weld joint penetration states based on multi-source data fusion. Firstly, a convolutional block, based on the architecture of convolutional neural networks, is designed. This forms part of an image representation network composed of three such blocks, incorporating a channel attention mechanism for high-quality image representation. Additionally, welding current and voltage data are integrated to create a comprehensive multi-source dataset. Secondly, an innovative atrous convolutional block is introduced, incorporating Efficient Channel Attention Networks (ECANet) to enhance the processing of multi-source data. Leveraging Temporal Convolutional Networks (TCN), an Efficient TCN (ETCN) with an advanced attention mechanism, is proposed. It is designed to extract global spatial features from the time series multi-source data, while concurrently feeding these data into a Transformer encoder to complete the extraction of time series features. Ultimately, a novel network utilizing a cross-attention mechanism is developed to amalgamate the time series and spatial features of the multi-source data, facilitating the prediction of Back-Side Bead Width (BSBW) and identification of the subsequent joint penetration state. Experimental findings indicate that irrespective of variations in welding current, the proposed algorithm achieves an RMSE of 0.17 mm. This represents a reduction in root mean squard error(RMSE), mean absolute error(MAE), and relative absolute error(RAE) values, along with an increase in R-square(R ^2 ) value, surpassing the performance of existing methods such as CNN, ResNet, CNN-LSTM, and AE-GRU.

The malicious mining pool can sacrifice part of its revenue to employ the computing power of blockchain network. The employed computing power carries out the pool mining attacks on the attacked mining pool. To realize the win-win game between the malicious mining pool and the employee, the paper proposes an Employment Attack Pricing Algorithm (EAPA) of mining pools in blockchain based on game theory. In the EAPA, the paper uses mathematical formulas to express the revenue of malicious mining pools under the employment attack, the revenue increment of malicious mining pools, and the revenue of the employee. It establishes a game model between the malicious mining pool and the employee under the employment attack. Then, the paper proposes an optimal computing power price selection strategy of employment attack based on model derivation. In the strategy, the malicious mining pool analyzes the conditions for the employment attack, and uses the derivative method to find the optimal utilization value of computing power, employees analyze the conditions for accepting employment, and use the derivative method to find the optimal reward value of computing power. Finally, the strategy finds the optimal employment computing power price to realize Nash equilibrium between the malicious mining pool and the employee under the current computing power allocation. The simulation results show that the EAPA could find the employment computing power price that realizes the win-win game between the malicious mining pool and the employee. The EAPA also maximizes the unit computing power revenue of employment and the unit computing power revenue of honest mining in malicious mining pool at the same time. The EAPA outperforms the state-of-the-art methods such as SPSUCP, DPSACP, and FPSUCP.