In the digital economy era, the transformation of all industries toward digitalization has become imperative. The digital transformation of standards is both a self-driven evolution and an intrinsic demand to adapt to the digital age. By establishing a comprehensive evaluation model for machine-readable standards and integrating it with application scenario-based testing platforms, a systematic assessment framework of “evaluation metrics-testing platform-machine-readable standard verification” has been developed. This helps to improve the stability and standardization of the quality of machine-readable standards, and promote the industrial application of machine-readable standards.
Financial time series prediction is crucial for unraveling the trends of financial indicators and aiding decision making. However, the main challenge is forecasting future trends from noise and non-linear historical data with low generability. Most existing methods involve model optimization for accuracy improvement but ignore the generability among multiple sequences, thereby limiting the forecast of high-dynamic time series. In this paper, we propose StockAgent, a multi-agent collaborative framework, to adaptively learn the multi-scale features from time series for stock price prediction. Specifically, StockAgent consists of a set of expert agents to perform parallel analyses of long- and short-term trends, periodic patterns, and risks. A summarizing manager agent is responsible for aggregating the analysis results and generating predictions, while a reflective manager agent performs optimization and adaptive weight adjustment of the expert agents based on experience feedback. We qualitatively and quantitatively evaluated the StockAgent on two datasets. Extensive experiments demonstrate that StockAgent is comparative to deep-learning based solutions in terms of prediction errors and exhibits outstanding zero-shot abilities.
The goal of biomedical relation extraction is to identify and extract high-quality information from extensive medical texts, which is crucial for advancing biomedical research. Recent advancements have led to significant progress in this field, with various methods demonstrating impressive results. Nevertheless, there remains substantial potential for further enhancement. In this paper, we introduce a Weighted Graph Convolutional Neural Network (WGCN) model designed specifically for extracting biomedical relations from medical texts. Our approach begins with obtaining the syntactic dependencies of the given sentences. We then integrate these dependencies with weighted information to improve relational extraction accuracy. Specifically, our method involves constructing an adjacency matrix and a dependency matrix based on the relations between words in the text, followed by the development of a corresponding weight matrix to account for the significance of different dependencies. This allows each dependency to contribute differently to the extraction of relation triples. Moreover, to address potential noise introduced by numerous dependencies within sentences, we implement a pruning strategy for the dependency tree. Experimental results on three publicly available biomedical datasets demonstrate that our proposed method is both effective and superior to the best-performing existing models. Further ablation studies confirm the effectiveness of each component within our model.
With the emergence of diversified mobile applications in the social field, e.g., image recognition, the massive growth in mobile data has imposing more service requests with strict real-time requirement on servers, therefore, the edge computing as an effective paradigm has been used to provide low-latency intelligence services. To improve the efficiency of concurrent processing on the edge server, the microservices architecture also has been adopted to offer a promising data processing solution. In this paper, through combining the stated advantages for social image recognition, an edge intelligence empowered scheme using microservices architecture is proposed. To enhance the efficiency of model training for image recognition, an optimized training model within the native framework of convolutional neural network (CNN) is firstly proposed. And then, to achieve the effective and low-latency social image recognition, a two-layer edge-end architecture is designed, within which, the image classification model is divided into feature extractor and image classifier, which are respectively deployed to the edge server and mobile terminals. Using the partial offloading, the mobile terminals perform the image preprocessing and classification, while the image feature extraction is executed on the edge server. In the experiments, the efficiency of the optimized training model is verified through comparing the training loss and accuracy, and the efficiency of the designed edge-end architecture is demonstrated with the concurrent testing.
Activity recognition for delivery personnel is growing in significance, as it empowers the quantification and evaluation of their performance by learning their operational behavior related to energy expenditure. This paper presents a deep learning model integrating LSTM and CNN, automating feature extraction and refining activity classification. LSTM captures temporal dependencies, while CNN handles spatial sensitivity, enabling the model to identify complex patterns. The model has demonstrated impressive accuracy, reinforcing its position as a tool for enhancing efficiency and well-being in logistics distribution.
Sparse Matrix-Matrix Multiplication (SpMM) is a widely used algorithm in Machine Learning, particularly in the increasingly popular Graph Neural Networks (GNNs). SpMM is an essential arithmetic operation in GNNs and has been parallelized on various platforms to accelerate GNN training. However, it has not been deeply studied on multi-GPU architectures.In this work, we parallelize the SpMM algorithm on MGPUSim, including the parallel execution mechanism and architecture design. More importantly, we propose an adaptive batching strategy (ABS) to handle the irregular memory access of sparse matrices and allocate work-item resources efficiently. The strategy addresses issues such as scheduling overhead caused by too large work-groups, resource contention, and poor parallelism resulting from performance degradation due to too small work-groups. ABS improves the GPU’s command processor scheduling speed, increases the efficiency of the incoming request rate, and optimizes work-group overhead. Finally, we conducted experiments on the i9-10900F CPU and NVIDIA RTX 3070 GPU using a set of matrices from SuiteSparse. Experimental results show that ABS achieves an average performance acceleration of 1.53× compared to the baseline and 1.40× compared to CUDA. Compared to the latest SpCaches approach used on FPGAs, our approach achieves a 1.12× relative performance improvement.
We are excited to introduce MetaScene, a ground-breaking framework for rendering Metaverse’s large-scale virtual 3D scenes that is versatile and lightweight. Previous construction solutions relied heavily on direct loading and rendering of 3D detailed models. However, our approach is simply to load a 3D simplified model and a live video stream projected onto the model. The 3D simplified model is derived from the 3D fine model, contains only the necessary bite, and the amount of data is much smaller. Real-time video streams are encoded with images rendered by a 3D fine model, and then the video stream is decoded and projected onto a 3D simplified model. Our MetaScene framework is so efficient that it can handle scenes of infinite scale, rendering quality on web pages is comparable to Unreal Engine (e.g. UE5), and can deliver photo-grade images. MetaScene supports cross-platform, such as web browsers, PC apps, mobile apps, and more. It also allows multiple users to interact seamlessly in virtual 3D scenes. And can be extended to any virtual 3D scene application. We have verified the data usage of multi-client parallel access through a large number of experiments, and our model can match or even exceed the UE5 framework in terms of FPS and bandwidth usage.
The human pose estimation network not only improves the prediction accuracy, but also brings the increase of network model parameters and computational complexity, so it is difficult to deploy in the practical application of devices with small computing power. Although lightweight human pose estimation method requires less processing and can be deployed on devices with fewer resources, its network model is not excellent. To solve these problems, a lightweight high-resolution human pose estimation network (GL-HRNet) based on attention mechanism is proposed. Specifically, GL-HRNet builds GCUup module by introducing global context module based on Small HRNet model. The basic module is redesigned with the structure of shuffling module. The cross-branch channel weighting operation is used to replace the computation-heavy point convolution, and the CBAM attention mechanism is used to enhance the module performance, and the SWCblock module is constructed. In addition, SCCtransition module is constructed by introducing CBAM attention mechanism reconstruction deep separable module to replace the transition layer module. The method was evaluated on the human body key point COCO dataset and compared with the current mainstream methods. Experimental results show that the proposed method can reduce the number of parameters and computational complexity, and improve the accuracy of the model effectively.
As a transparent object, glass presents a huge challenge to computer vision tasks, but glass is inevitably present in every corner of our lives. As a new image saving method, panoramic image can save more semantic information and is widely used in various scenes. Therefore, it is very meaningful to study glass segmentation based on panoramic images. However, glass segmentation based on panoramic images not only needs to overcome the characteristics of transparent reflection of glass, but also needs to overcome the distortion of panoramic images. To solve this problem, our team proposed a new network architecture, the Panoramic SegGlass network. In order to solve the difficult problem of panoramic image glass segmentation, we combine the spatial deformable projection embedding, deformable multi-layer perceptron, self-focusing mechanism, fusion operation and channel-level multi-layer perceptron, and propose two modules GlobalLocalFusion and ChannelDeformPercept. Through a large number of experiments, our network structure has achieved Iou, MAE and Fscore of 90.62, 0.0073 and 94.91, respectively, surpassing the existing model. It also shows that our model is competitive on other glass datasets.
In response to prevalent challenges in current surface defect detection methods for steel, such as high algorithm complexity, extensive computational requirements, difficulty in extracting features from intricate defect images, and low detection accuracy, a lightweight steel surface–defect detection method based on the YOLOv8n target-detection model incorporating cross-scale feature fusion was proposed in this paper. First, we introduced a lightweight convolutional neural network-based cross-scale feature fusion module to enhance the neck network in the YOLOv8n model. This modification effectively reduced the model’s parameter count and computational load. Second, to increase the model’s detection precision through the efficient extraction of local detailed features of defect images, we integrated an inverted residual mobile block into the C2f module of the YOLOv8n model. Finally, implementing the shape intersection over union loss function improved the model’s performance while accelerating the convergence speed. Experimental results obtained from the publicly available steel dataset NEU-DET showcased notable improvements. Compared with the original model, the enhanced YOLOv8n model reduces parameters and giga floating point operations per second by 30.2% and 12.3%, respectively. Moreover, this improved model achieved precision and mean average precision of 80.5% and 80.3%, respectively, indicating enhancements of 5.8% and 1.0% over the original model. These advancements considerably enhance detection effectiveness while ensuring model lightweightness.
As the rapid development of communication technology, more and more smart devices will be connected in cognitive radio Internet of Things (CIoT) using an opportunistic manner. However, the dynamicity and heterogeneity of spectrum will make it more difficult in data transmission strategy design for unreliable IoT. Opportunistic routing exploits the broadcast property of wireless medium to improve network performance in an opportunistic communication manner. Inspired by this, in this paper, we propose a Nash-learning and Transfer Learning based Opportunistic Data Transmission scheme, NTLODT, for differentiated services CIoT, which exploits Nash-learning for candidate set optimization and transfer learning for data transmission decision from a trust mechanism perspective. In addition, we prove that the convergence and complexity of the proposed algorithm. The simulation results demonstrate that the effectiveness of NTLODT in terms of average energy cost per bit, average delay, expected cost of routing and throughput.
In beyond 5G (B5G) Internet of Things (IoT) system based on edge intelligence, pay-for-use demand has become a consensus, and the pricing of IoT services has attracted the attention of academia and industry. The pricing method based on noncooperative game allows edge service providers (ESPs) to compete fairly, effectively preventing edge nodes from malicious bidding. However, since only one winner can make a profit each time, it is easy to cause a large number of ESPs to lose money for a long time. To this end, a dynamic alliance pricing mechanism based on distributed edge intelligence is proposed. ESPs can freely choose to form an edge dynamic alliance, which not only retains the independence of edge nodes but also makes full use of the advantages of mutual cooperation between nodes. According to the characteristics of edge nodes, various roles are reasonably divided. In order to prevent abnormal behaviors of edge nodes, we set up necessary restrictive rules. At the same time, we designed a privacy-enhanced joint pricing prediction algorithm to screen the alliance's candidate solutions to improve pricing efficiency and edge benefit. The experimental results show that, compared with the traditional alliance game method, the performance of the mechanism we proposed improves the utilization rate of edge resources by 32.76%-61.37%. Meanwhile, the prediction accuracy was improved by 16.47%-38.86%, and the average prediction time was reduced by 42.81%-65.57%.
Binary code similarity detection (BinSD) systems trend to utilize deep learning to identify semantic features of assembly code and exhibits superior performance, gaining increasing popularity against traditional methods. However, it has been observed that existing deep learning models are susceptible to data poisoning attacks, posing a latent threat to the robustness and reliability of BinSD. Existing data poisoning strategies in BinSD are easily detectable for the generated triggers will destroy code functions. Moreover, selecting trigger injection location needs repeated exploration and verification, increasing the attack cost. To address this issue, we propose a novel adversarial scheme, named as AdvBinSD, which can poison the deep learning-based binary code similarity detector and make it sensitive to isolated instruction sequences. In AdvBinSD, the isolated instruction sequences generally refer to those instructions that have no data dependencies with other instructions and do not affect the function of original binary code, and also it is difficult to discovery those isolated instruction sequences by verifying syntactic validity and semantic integrity. Different from existing data poisoning strategies, AdvBinSD first estimates a code fragment that has the greatest impact on software functionality as the poisoning location, and then add isolated instruction sequences into this location to synthesize effective poisoned samples. This location estimation is achieved by maximizing the similarity between function-level feature vectors and instruction-level feature vectors, ensuring that the modified assembly code can execute correctly. Furthermore, to improve the efficiency of feature vector similarity computing process, a k-order greedy feature comparison (k-GFC) algorithm is also designated. Extensive experiments demonstrate that our proposed AdvBinSD can successfully poison the state-of-the-art deep learning-based binary code similarity detectors.
Enterprise networks are divided into sub-domain structures for resource aggregation and trust dependency. However, the complex domain structure results in delayed and incomplete defenses from Intrusion Response Systems(IRS). This paper proposes a proactive intrusion response system specifically designed for multi-domain networks called ProMD. A Bayesian Risk Graph is designed to analyze and evaluate the risks dynamically. The paper formalizes the selection of countermeasures as a multi-objective optimization problem (MOOP) and proves it is NP-Hard. Therefore a genetic-based heuristic algorithm OPGA is designed to generate optimal response policy sets and an unique policy is selected based on the Entropy Weight method. Experimental results demonstrate that the OPGA outperforms benchmark algorithms.
Bit-Flip Attack (BFA), an adversarial attack on network parameters, has proven to be a great threat to the security of Deep neural networks (DNNs). More concerning, recent works introduce a variant of BFA called Targeted Bit-Flip Attack (T-BFA), a targeted and stealthy attack paradigm. DNNs under T-BFA misclassify targeted inputs and behave normally on untargeted inputs, which makes the attack undetectable for users. Although existing defense methods can mitigate the effect of T-BFA, their lack of consideration for T-BFA characteristics results in unnecessary accuracy loss or run-time overhead. In this study, we leverage the “targeted” characteristic of T-BFA and introduce a two-step defense method to enhance the robustness of the DNN against T-BFA more cost-effectively. Firstly, in order to protect the security-sensitive classes that we identified based on T-BFA preferences from misclassification, we propose a method to locate security-sensitive weights crucial for classification of security-sensitive classes in convolutional and fully-connected layers. Secondly, in order to mitigate the effect of T-BFA and achieve a better trade-off between robustness and accuracy. We propose Random Switch to reduce the expected impact of bit flips on weights, and a selection strategy based on the benefit and negative impact of protecting a weight with Random Switch to further select security-sensitive weights for deployment of Random Switch. Experimental results show that our proposed method effectively enhances the robustness of DNNs against TBFA. Compared with the baseline, our method can tolerate more than $5.1 \times $ flipped bits with negligible accuracy loss (< 0.1% in resnet-50).
In various applications in Internet of Things like industrial monitoring, large amounts of floating-point time series data are generated at an unprecedented rate. Efficient compression algorithms can effectively reduce the size of data, enhance transmission performance and storage efficiency, and simultaneously lower storage costs. Therefore, there is a need for lightweight and efficient stream compression algorithms. In this paper, we propose a novel lossless floating-point data compression algorithm called Ant. The main idea is to encode double-precision floating-point numbers into integer form, calculate the delta between adjacent values, and then convert the delta into unsigned integers. This encoding method effectively reduces storage costs and improves data compression efficiency. Extensive experiments on real-world datasets demonstrate that our algorithm achieves compression speeds at least as fast as state-of-the-art streaming methods, and a 63% relative improvement in average compression rate.
Location-Based Skyline Query (LSQ) is a computational method aimed at retrieving non-dominated data points from spatial datasets. As data owners have limited resources, it is common for them to outsource data to clouds and provide query services for others. But this raises privacy challenges. Existing privacy-preserving LSQ schemes lack practicality due to computational or spatial overheads, with some requiring the data owner to perform heavy computations. To address this, we propose an innovative privacy-preserving LSQ scheme ensuring both data owner and query user privacy. We design a privacy-preserving point-convex intersection determination scheme (PPID) that enables a cloud server to determine whether a point intersects a polygon without knowing the real data. Based on PPID, we devise an efficient privacy-preserving LSQ scheme (PLSQ). Security analysis demonstrates PPID and PLSQ are safe in real/ideal world model security, and PLSQ can preserve dataset and query privacy. Extensive experiments reveal PLSQ’s efficiency, achieving search speed enhancements of $4 - 7 \times $ and decreasing data owner overhead by 25% – 40% compared to related schemes.
Vehicle-to-vehicle (V2V) energy trading stands as a significant technology, allowing electric vehicles (EVs) to share energy. This balances energy demand and supply, reducing pressure on the power grid. However, two substantial challenges require attention: 1) Creating a standardized protocol universally adoptable by EV manufacturers and energy trading participants to establish a common secure and reliable platform for real-time energy trading between EVs. 2) Determining a fair and transparent pricing and transaction management mechanism, accounting for factors such as energy demand and supply, location, and user preferences. To tackle these hurdles, this paper proposes a decentralized V2V energy trading system (VETS) based on a novel high-performance sharding blockchain framework. VETS incorporates a standard energy trading management protocol, achieving trustless and transparent transaction processing. It employs a blockchain sharding and parallel architecture for high transaction throughput, catering to a large number of EVs. Furthermore, an efficient consensus algorithm is developed to minimize service delays and adapt to real-time transaction demands. A practical testbed operationalizes VETS, facilitating experimental analysis within a real-world cloud environment.
This paper proposes a talking face generation method named "CP-EB" that takes an audio signal as input and a person image as reference, to synthesize a photo-realistic people talking video with head poses controlled by a short video clip and proper eye blinking embedding. It's noted that not only the head pose but also eye blinking are both important aspects for deep fake detection. The implicit control of poses by video has already achieved by the state-of-art work. According to recent research, eye blinking has weak correlation with input audio which means eye blinks extraction from audio and generation are possible. Hence, we propose a GAN-based architecture to extract eye blink feature from input audio and reference video respectively and employ contrastive training between them, then embed it into the concatenated features of identity and poses to generate talking face images. Experimental results show that the proposed method can generate photo-realistic talking face with synchronous lips motions, natural head poses and blinking eyes.
The rapid development of intelligent transportation technology has promoted the progress of multiple trains cooperative technology. This paper proposes an online cooperative cruise control method based on improved particle swarm optimization (IPSO) and deep feedforward neural network (DFNN), which enables each train to adaptively adjust the controller according to its own state information, ensuring speed synchronization and maintaining safe distance intervals between trains. Through the collection of information from the train’s big data, the optimal time-varying cooperative control gain is fitted to achieve efficient, safe, and real-time control of multiple high-speed trains.
