The rapid proliferation of the Internet of Things (IoT) poses significant challenges for utility optimization in sharding blockchain systems. In this paper, we propose a Committee Scheduling Algorithm (CSA), which employs an iterative optimization framework based on the Markov chain to balance transaction throughput, cumulative latency, and transaction fees. CSA dynamically adjusts the committee members to achieve near-optimal solutions while addressing operational constraints. Theoretical analysis demonstrates the convergence bounds of the algorithm and its robustness against Sybil and eclipse attacks, ensuring high entropy for committee selection. Experimental results show that CSA outperforms Stochastic-Exploration (SE), Simulated Annealing (SA), and Policy Gradient-Based Computing Task Scheduling (PG-CTS) in terms of utility, convergence speed, and adaptability to dynamic events, with the committee scheduling utility improving by about 30%. Furthermore, CSA demonstrates stable performance in large-scale IoT environments characterized by dynamic node additions and failures. This paper offers a robust and adaptive solution for utility optimization in sharding blockchains, thereby improving the scalability, security, and efficiency of IoT applications.

Multimodal sentiment analysis faces dual challenges of asynchronous audio-visual fusion and robustness to modality absence. To address these, this paper proposes a Hierarchical Transformer with Dual-level Feature Enhancement (HTMD). The framework first employs mutual promotion units to dynamically calibrate asynchronous audio-visual streams: audio features are optimized through an adaptive time-frequency attention mechanism, while video representations are enhanced via a multi-scale local-global attention mechanism. For modality absence issues, the framework implements feature reconstruction complemented by a modified MAE loss function to ensure training stability. Comprehensive evaluations on CMU-MOSI and CH-SIMS datasets demonstrate that HTMD significantly outperforms existing methods in both complete and partial modality scenarios, establishing new benchmarks for robust multimodal sentiment analysis.

This paper presents a Multimodal Dynamic Inference Framework (MDIF) for video question answering in traffic scenarios. MDIF is different from existing methods that rely on correlation modeling and global features. It combines causal inference with object segmentation. Causal inference reduces spurious correlations and ensures unbiased reasoning. Object segmentation extracts fine-grained information such as traffic signs, vehicle movements, and pedestrian interactions. Together, they improve causal modeling, scene understanding, and event prediction accuracy. MDIF also employs Graph Convolutional Networks (GCNs) to capture spatiotemporal dependencies in complex and dynamic traffic events. We evaluate MDIF on two large-scale datasets, SUTD-TrafficQA and MSVD-QA. The results demonstrate significant progress in prediction and counterfactual tasks, validating the robustness and generalization ability of the MDIF framework. Ablation studies further confirm the role of spatiotemporal modeling, object segmentation, and causal intervention. These components are all essential to robust, interpretable, and unbiased reasoning. Overall, MDIF provides strong advantages in causal modeling and cross-modal inference.

The detection of small objects in unmanned aerial vehicle (UAV) images remains a significant challenge due to the constraints imposed by UAV altitude and shooting angles. Objects in such images are often small and lack distinct features, making them vulnerable to interference from lighting variations, shadows, and background noise, which complicates detection tasks. To address these challenges, we propose an attention-based fusion network with an upsampler-free feature pyramid network (AUFFNet), designed to enhance the detection of small objects in UAV aerial imagery. The proposed framework incorporates three key innovations: the additive self-attention convolutional gated fusion module, which enhances multi-dimensional feature interactions to improve small object representation while mitigating background interference; the inter-layer feature pyramid network in UAV aerial images, an upsampler-free feature fusion architecture that avoids feature loss caused by traditional upsampling and downsampling operations through cross-scale feature fusion; and the remote sensing dense small target detection loss function, which further reduces prediction errors by addressing issues related to occlusion and scale variations. Extensive experiments on the VisDrone dataset demonstrate the effectiveness of AUFFNet. Compared with the baseline YOLOv10s model, AUFFNet achieves significant improvements in performance metrics, including precision of 57.6% (+5.9%), recall of 49.8% (+11.2%), mAP50 of 48.8% (+8.7%), and mAP50-95 of 28.7% (+4.2%). Furthermore, the model achieves these results with a reduced parameter count of 5.6M (- 2.4M), highlighting its efficiency and practicality for real-world UAV applications. These advancements underscore the potential of AUFFNet as a robust solution for small object detection in complex aerial environments.

There is a need for information transactions between nodes in the Internet of Things (IoT) and blockchain technology can guarantee the anonymity and security of such transactions. However, current blockchain systems require networks to be connected in real time, but IoT networks cannot fulfill this requirement. Therefore, we put forward a blockchain parallel activity architecture using carriers (CBPA), which is capable of deploying blockchain systems on IoT networks. Firstly, the blockchain operation architecture, and its components are demonstrated. Secondly, the generation methods of the carrier, the carrier block and the transaction block are designed, respectively. Additionally, a feature-to-transaction correspondence algorithm is proposed, with the objective of accommodating the previous work of nodes when they are in a disconnected state within the network. Thirdly, a parallel generation method for transaction blocks is designed to permit multiple nodes to collaborate in generating blocks, thereby reducing the difficulty of block generation while accelerating the generation speed. Finally, intra-block and cross-block conflict resolution algorithms, as well as a block consensus and fork processing algorithm, are designed to ensure that nodes can participate in blockchain activities without being at a disadvantage and obtain legal benefits even when operating on a network with high communication latency. Theoretical analysis indicates that CBPA has both security and liveness. The experimental results show that when the block size is 1MB, CBPA improves the average throughput by about 10% and reduces the average latency by about 14% compared to existing schemes. When the percentage of failed nodes reaches about 18%, the blocking time increases significantly, and the valid block rate decreases by 0.2%. The proposed CBPA architecture expands the applicable blockchain network and provides practical solutions for disconnected operations.

As technology advances, autonomous driving systems are gradually maturing. However, in complex outdoor conditions such as heavy rain or dense fog, these systems often face challenges due to low visibility, high noise levels, and loss of detail. To address these issues, this paper introduces an improved method for infrared target detection based on YOLOv8n. Initially, a FasterNet Block module is incorporated to enhance computational speed and the efficiency of spatial feature extraction. Subsequently, the MACSP module was introduced in the network’s neck. This module enhances the efficiency of feature extraction from images, enabling the capture and effective utilization of multi-scale features. Furthermore, enhancements to the detection head through the WiseIoU loss function have improved target localization accuracy. To demonstrate the effectiveness and versatility of the proposed method, experiments were conducted on the updated FLIR (FLIR_ADAS_v2) infrared dataset and the RTTS fog dataset. The experimental results reveal that the FMA-YOLO algorithm achieves an increase in mean Average Precision (mAP) by 2.0

This study aims to give a comprehensive overview of the application of the metaverse in educational evaluation. First, we characterize the metaverse and illustrate how it can support educational evaluation from the perspectives of virtual reality, augmented reality, and blockchain. Then, we outline the metaverse exploration framework and summarize its technical advantages. Based on this, we propose a metaverse-based implementation scheme to address the issues of reliability, accuracy, and credibility in educational evaluation. Finally, we show its implementation difficulties, performance evaluation, and future work. This proposed scheme opens up new research directions for the reform of educational evaluation while expanding the potential and reach of metaverse applications in education. We think that this study can help researchers in building an ecosystem for educational evaluation that is trustworthy, equitable, and legitimate.

In the digital information age, distributed file storage technologies like the InterPlanetary File System (IPFS) have gained considerable traction as a means of storing and disseminating media content. Despite the advantages of decentralized storage, the proliferation of decentralized technologies has highlighted the need to address the issue of file ownership. The aim of this paper is to address the critical issues of source verification and digital copyright protection for IPFS image files. To this end, an innovative approach is proposed that integrates blockchain, digital signature, and blind watermarking. Blockchain technology functions as a decentralized and tamper-resistant ledger, recording and verifying the source information of files, thereby establishing credible evidence of file origin. A digital signature serves to authenticate the identity and integrity of the individual responsible for uploading the file, ensuring data security. Furthermore, blind watermarking is employed to embed invisible information within images, thereby safeguarding digital copyrights and enabling file traceability. To further optimize the efficiency of file retrieval within IPFS, a dual-layer Distributed Hash Table (DHT) indexing structure is proposed. This structure divides file index information into a global index layer and a local index layer, significantly reducing retrieval time and network overhead. The feasibility of the proposed approach is demonstrated through practical examples, providing an effective solution to the copyright protection issues associated with IPFS image files.

Current blockchain systems have high requirements on network connection and data transmission rate, for example, nodes have to receive the latest blocks in time to update the blockchain, nodes have to immediately broadcast the generated block to other nodes for consensus, which restricts the blockchain to run only on real-time connection networks, but the existence of delay tolerant networks poses a great challenge to the deployment of blockchain systems. To address this challenge, a novel blockchain transaction mechanism is proposed. First, the block structure is modified by adding a flag, and on this basis, the definition of the extrachain is proposed. Secondly, based on the blockchain transaction process, transaction verification and consensus algorithms on the extrachain are presented. Thirdly, both the extrachain selection algorithm and appending algorithm are proposed, so that the extrachain can be appended to the blockchain fairly and safely. Finally, an extrachain transmission scheme is presented to broadcast the blocks generated in the delayed network to the normal network. Theoretical analysis and simulation experiments further illustrate the efficiency of the proposed mechanism.

Currently, there is a great deal of interest in multimodal aspect-level sentiment classification using both textual and visual information, which changes the traditional use of only single-modal to identify sentiment polarity. Considering that existing methods could be strengthened in terms of classification accuracy, we conducted a study on aspect-level multimodal sentiment classification with the aim of exploring the interaction between textual and visual features. Specifically, we construct a multimodal aspect-level sentiment classification framework with multi-image gate and fusion networks called MFSC. MFSC consists of four parts, i.e., text feature extraction, visual feature extraction, text feature enhancement, and multi-feature fusion. Firstly, a bidirectional long short-term memory network is adopted to extract the initial text feature. Based on this, a text feature enhancement strategy is designed, which uses text memory network and adaptive weights to extract the final text features. Meanwhile, a multi-image gate method is proposed for fusing features from multiple images and filtering out irrelevant noise. Finally, a text-visual feature fusion method based on an attention mechanism is proposed to better improve the classification performance by capturing the association between text and images. Experimental results show that MFSC has advantages in classification accuracy and macro-F1.

The essence of blockchain is the competition for node mining rights,and the real-time connectivity of the network is prerequisite and guarantee.However,there exist delay-tolerant networks with data transmission latency,which puts nodes at a disadvantage in blockchain activities.To address this problem,a transaction architecture is constructed to provide service support for nodes in discontinuous connectivity networks to participate in blockchain activities.Firstly,an auxiliary block named as an attached block is constructed as a basis for distinguishing new blocks generated when the network is connected or disconnected.Secondly,with the miner's identity and several random numbers as parameters,unique and verifiable discrete tokens are generated to obtain mining qualifications.On this basis,the mining qualification attribution algorithm and branch processing algorithm are designed.The for-mer avoids the problem of the richest man doing evil in PoS(proof of stake)and DPoS(delegated proof of stake),and the latter ensures the attached chains are added to the main chain with an equal probability.Finally,a quadratic consensus scheme is designed to solve the problem of false and duplicate transactions that may exist in generated blocks when the network is disconnected.Theoretical proof and experimental analysis show that the architecture has advantages in indices such as reliability,acceptability,transaction throughput,confirmation time and branching rate.

With the advent of the fourth industrial revolution, data-driven decision making has also become an integral part of decision making. At the same time, deep learning is one of the core technologies of the fourth industrial revolution that have become vital in decision making. However, in the era of epidemics and big data, the volume of data has increased dramatically while the sources have become progressively more complex, making data distribution highly susceptible to change. These situations can easily lead to concept drift, which directly affects the effectiveness of prediction models. How to cope with such complex situations and make timely and accurate decisions from multiple perspectives is a challenging research issue. To address this challenge, we summarize concept drift adaptation methods under the deep learning framework, which is beneficial to help decision makers make better decisions and analyze the causes of concept drift. First, we provide an overall introduction to concept drift, including the definition, causes, types, and process of concept drift adaptation methods under the deep learning framework. Second, we summarize concept drift adaptation methods in terms of discriminative learning, generative learning, hybrid learning, and others. For each aspect, we elaborate on the update modes, detection modes, and adaptation drift types of concept drift adaptation methods. In addition, we briefly describe the characteristics and application fields of deep learning algorithms using concept drift adaptation methods. Finally, we summarize common datasets and evaluation metrics and present future directions.

In recent years, the application of deep learning in many fields has gradually increased, and the practice in the field of information recommendation has also achieved excellent consequence. DeepFM model is used to share input data between wide and deep parts, it does not need artificial feature engineering, which can handle the advantages of low and high-order features simultaneously. In this paper, the hidden layer of DeepFM model is improved to solve the problem of gradient vanishing and gradient explosion in deep neural network effectively, and the improved algorithm model is verified by experiments on the MovieLens dataset. The improved model achieved a 1.29% increase in AUC indicator, and reduced the deviation between the predicted value and the true value of the model to a certain extent. Practice has proved that the improved model solves the problem of sparse film user rating data in the movie recommendation scenario, the accuracy of movie recommendation has been improved indirectly.

基于元宇宙虚实融合、多维交互、非中心化信任体系等典型特征,分析总结元宇宙赋能学习评价改革的技术优势,探讨新技术加持下的评价实现途径,提出可靠、科学和可信的应用架构,包括元宇宙支撑下的评价数据采集、激励驱动的评价数据处理和扩展驱动的评价数据存储,以有效解决当前学习评价存在的采集数据单一、评价方法不合理、存储数据不可信等缺陷.

元宇宙为新工科教育发展提供了重要机遇,其与教育领域的融合成为破解当前新工科教育改革的有效手段.本文首先总结了元宇宙赋能新工科教育的主要特征,然后提出了元宇宙在新工科教育的应用架构,其中包括情境化教学环境构建、多维化教学方式互动、可信化教育数据保障和共享化教育资源创建,最后分析了应用架构所面临的技术挑战.本文的研究成果为加快推进元宇宙与新工科教育相结合提供了有益参考.

Blockchain proposes many innovative technologies to establish credible mechanisms in an open environment and therefore, it becomes a promising solution to the problem of credibility in educational development. To better understand the role of the blockchain, we aim to provide an extensive survey focusing on its key technology, application potential, and performance evaluation. First, from the perspective of blockchain characteristics, we summarize its application architecture in educational credibility. Next, we extensively discuss application potential of the blockchain, such as data storage, data sharing, achievement certification, and activity evaluation. Moreover, we investigate the performance evaluation, including basic performance metrics and specialized metrics for credibility. Finally, we analyze the challenges and research trends of blockchain in educational credibility and provide useful insights for future research.

In this paper, we study the application of Metaverse in learning evaluation, focusing on in-depth exploration from the perspective of extended reality and blockchain. First, we present a metaverse exploration framework, summarizing the role of its technological advantages in learning evaluation. On this basis, we propose a solution scheme to the problems of reliability, scientificity, and credibility in learning evaluation, and discuss the key issues of the solution. We believe that this paper can provide useful insights into the field of metaverse exploration, and open up new ideas for learning evaluation.

Saliency object detection is a fundamental problem in the field of computer vision. With the commercial success of consumer-grade depth sensors such as Microsoft Kinect, the captured RGB-D images provide users with a higher viewing experience, but also pose a higher challenge to the current saliency detection technology. In this paper, we propose a shape-aware saliency object detection approach SASD, manifesting in two aspects: 1) obtaining high quality saliency maps, especially the salient details; 2) displaying irregular salient regions, such as circles, ellipses, and splines. The proposed SASD approach consists of three steps. First, the initial saliency map is obtained by combing depth information and edge details. On this basis, the rules for the dominant factors of saliency detection are designed and the method of enhanced saliency map calculation is proposed. Finally, an irregular shape display method is demonstrated, the purpose of which is to fit the obtained saliency map to different shapes. The experiment demonstrates the effectiveness of our approach in subjective and objective aspects.

The current blockchain deployment solutions rely on a continuous connectivity network. Unfortunately, the delay tolerant network does not meet this condition. Therefore, we construct a novel blockchain transaction framework with discrete token negotiation called DTNB, which can be deployed on the delay tolerant network to provide transaction services. Specifically, we present the structure of add-chains by modifying the existing block structure. Then, we design a mining qualification determining scheme to achieve fair transactions, including discrete token generation algorithm and mining qualification attribution algorithm, and this scheme avoids the problem that nodes with more stakes in the PoS and DPoS algorithms can obtain mining qualifications with a higher probability. Furthermore, we present two mining schemes and also design a fork processing algorithm, which ensures that blocks on the add-chain generated by the local network can be appended to the main chain with the equal probability. Finally, we design the second consensus algorithm to avoid the problem of false and repeated transactions of the blocks on the add-chains in the local network. Theoretical analysis shows three properties of DTNB, including safety, reliability and activeness, and the experimental simulations demonstrate DTNB has advantages in throughput, block generation time and fork rate.

为了讨论配煤煤质和焦炭强度间的统计学关联性,选取配煤的煤质指标:灰分(Ad)、挥发分(Vdaf)、全硫(Std)、黏结指数(G)等为 自变量,以焦炭的强度指标:抗碎强度(M40)、耐磨强度(M10)、反应性指数(CRI)以及反应后强度(CSR)为因变量,采用SPSS软件对配煤煤质参数与焦炭强度相关性进行了多元线性回归分析,并完成了预测模型的构建.研究结果表明,此方法迅速且较为精确,适于焦炭质量预测和控制.