Accurate and efficient implementation of parallel quantum gates is crucial for scalable quantum information processing. However, the unavoidable crosstalk between qubits in current noisy processors impedes the achievement of high gate fidelities and renders full Hilbert-space control optimization prohibitively difficult. Here, we overcome this challenge by reducing the full-system optimization to crosstalk-robust control over constant-sized subsystems, which dramatically reduces the computational cost. Our method effectively eliminates the leading-order gate operation deviations induced by crosstalk, thereby suppressing error rates. Within this framework, we construct analytical pulse solutions for parallel single-qubit gates and numerical pulses for parallel multi-qubit operations. We validate the proposed approach numerically across multiple platforms, including coupled nitrogen-vacancy centers, a nuclear-spin processor, and superconducting-qubit arrays with up to 200 qubits. As a result, the noise scaling is reduced from exponential to linear for parallel single-qubit gates, and an order-of-magnitude reduction is achieved for parallel multi-qubit gates. Moreover, our method does not require precise knowledge of crosstalk strengths and makes no assumption about the underlying qubit connectivity or lattice geometry, thereby establishing a scalable framework for parallel quantum control in large-scale quantum architectures.

Quantum metrology promises to surpass classical precision limits by leveraging quantum resources such as entanglement. Maximally entangled Greenberger-Horne-Zeilinger (GHZ) states are theoretically optimal probes for quantum metrology. However, they are fragile to environmental noise, severely limiting their practical utility. To overcome this limitation, we propose and experimentally realize a noise-adaptive quantum metrology scheme that autonomously identifies optimal probes without any prior information of the noise. This is achieved by combining a variational quantum circuit that optimizes available resources with an efficient strategy for evaluating the sensing performance. Using a seven-qubit nuclear spin sensor, we identify optimal probe states that achieve precision improvements up to 0.698 dB over GHZ states when sensing a fixed magnetic field. Furthermore, we show that the learned probe state exhibits cross-parameter robustness, maintaining superior performance over a broad range of magnetic field frequencies. The proposed scheme is model-free, hardware-efficient, and scalable, providing a practical route to noise-resilient quantum metrology on near-term quantum devices.

Distributed learning is commonly used for training deep learning models, especially large models. In distributed learning, manual parallelism (MP) methods demand considerable human effort and have limited flexibility. Hence, automatic parallelism (AP) methods have recently been proposed for automating the parallel strategy optimization process. Existing AP methods suffer from sub-optimal solutions because they do not jointly optimize the two categories of parallel strategies (i.e., inter-layer parallelism and intra-layer parallelism). In this paper, we propose a novel AP method called UniAP, which unifies inter- and intra-layer automatic parallelism by mixed integer quadratic programming. To the best of our knowledge, UniAP is the first parallel method that can jointly optimize the two categories of parallel strategies to find an optimal solution. Experimental results show that UniAP outperforms state-of-the-art methods by up to 3.80× in throughput and reduces strategy optimization time by up to 107× across five Transformer-based models.

Digital signal processor (DSP) array has been widely used in deep learning accelerators. DSP packing is always used to reduce the chip area of DSP arrays. Weight approximation is typically adopted in DSP packing to further reduce chip area. However, existing methods use indiscriminate weight approximation (IWA), which will lead to significant accuracy decrease for large language models (LLMs) and high lookup table (LUT) consumption for LLM accelerators. In this paper, we propose a new method, called discriminate weight approximation (DWA), for efficient DSP packing in LLM accelerators. DWA can reduce LUT consumption while guaranteeing accuracy. Experimental results show that compared with methods without weight approximation, DWA achieves a 1.5 times reduction of chip area and an imperceptible accuracy decrease. Compared with existing IWA methods, DWA achieves the same chip area reduction performance but reduces the LUT consumption by 2.6 to 4.6 times, which results in a five times reduction of deployment cost. Moreover, the accuracy of DWA is far better than that of existing IWA methods.

Entanglement plays a crucial role in advancing quantum technologies and exploring quantum many-body simulations. Here, we introduce a protocol aided by neural networks for measuring entanglement in both equilibrium and non-equilibrium states of local Hamiltonians, with a favorable amount of training data. Our numerical simulations across various Hamiltonian models and qubit configurations reveal that this approach can predict comprehensive entanglement metrics, such as Rényi entropy, for up to 100 qubits using only single-qubit and two-qubit Pauli measurements. Excitingly, future entanglement dynamics beyond the measurement window can be predicted based solely on previous single-qubit traces. Experimentally, we utilize a nuclear spin quantum processor and a neural network to measure entanglement in the ground and dynamical states of a one-dimensional spin chain. The results demonstrate the feasibility of our method in practical experiments. Therefore, our approach offers a promising method for experimentally measuring entanglement in systems with dozens to hundreds of qubits.

With the rapid development of autonomous driving technology, traffic sign recognition (TSR) has emerged as a foundational component of mobile driving systems. Although significant progress has been made in current research, existing techniques still face challenges in recognizing traffic signs under complex weather conditions. This model employs an attention-based dynamic sequence fusion feature pyramid, which enhances recognition accuracy for small-target traffic sign instances in adverse weather, as opposed to traditional feature pyramid networks. Additionally, the model integrates a dynamic snake convolution operator along with Wise-IoU, enabling it to capture fine small-scale feature information while mitigating the impact of low-quality instances. Furthermore, the model introduces a novel data augmentation library, Albumentations, to simulate real-world complex weather scenarios, and utilizes a new performance evaluation metric, TIDE, to more effectively assess model performance in such conditions. We demonstrate the effectiveness of our model on the TT-100 K dataset, the GTSDB dataset, and the BDD 100 K dataset, achieving improvements in mAP of 9%, 1.5%, and 2.6%, respectively. Compared to the baseline model, Cls and Loc metrics decreased by approximately 3 and 1.2.The experiments indicate that our model exhibits excellent generalization ability and robustness, successfully performing small target detection under complex weather conditions in the realm of traffic sign recognition.

The high memory and computation demand of large language models (LLMs) makes them challenging to be deployed on consumer devices due to limited GPU memory. Offloading can mitigate the memory constraint but often suffers from low GPU utilization, leading to low inference efficiency. In this work, we propose a novel framework, called pipelined offloading (PIPO), for efficient inference on consumer devices. PIPO designs a fine-grained offloading pipeline, complemented with optimized data transfer and computation, to achieve high concurrency and efficient scheduling for inference. Experimental results show that compared with state-of-the-art baseline, PIPO increases GPU utilization from below 40% to over 90% and achieves up to 3.1$\times$ higher throughput, running on a laptop equipped with a RTX3060 GPU of 6GB memory.

The viability of the star count (Wolf) method is assessed as a means of constraining the near-infrared (NIR) extinction law toward the Corona Australis molecular cloud. Using deep JHK S photometry from the VISIONS survey, extinction maps with 1′ spatial resolution are constructed. The derived extinction ratios are A J / A H = 1.73 ± 0.07, A H / A K S = 1.70 ± 0.11 , and A J / A K S = 3.02 ± 0.22 , which are consistent with Galactic literature means. Assuming a power-law form ( A λ ∝ λ − α ) for the NIR extinction law, we derive indices of α ≈ 2.0 across all wavelength combinations, with no statistically significant wavelength dependence throughout the NIR wavelength range. While spatial variations in extinction properties are tentatively observed across the cloud, concerns persist regarding the impact of photometric completeness, and the role of reference field selection. Continued research is required to refine the approach, and scrutinize the veracity of potential extinction law variations over a more expansive region of sky.

We present a robust quantum optimal control framework for implementing fast entangling gates on ion-trap quantum processors. The framework leverages tailored laser pulses to drive the multiple vibrational sidebands of the ions to create phonon-mediated entangling gates and, unlike the state of the art, requires neither weak-coupling Lamb-Dicke approximation nor perturbation treatment. With the application of gradient-based optimal control, it enables finding amplitude- and phase-modulated laser control protocols that work without the Lamb-Dicke approximation, promising gate speeds on the order of microseconds comparable to the characteristic trap frequencies. Also, robustness requirements on the temperature of the ions and initial optical phase can be conveniently included to pursue high-quality fast gates against experimental imperfections. Our approach represents a step in speeding up quantum gates to achieve larger quantum circuits for quantum computation and simulation, and thus can find applications in near-future experiments.

The existing deep learning-based detection of fake information focuses on the transient detection of news itself. Compared to user category profile mining and detection, transient detection is prone to higher misjudgment rates due to the limitations of insufficient temporal information, posing new challenges to social public opinion monitoring tasks such as fake user detection. This paper proposes a multimodal aggregation portrait model (MAPM) based on multi-model joint representation for social media platforms. It constructs a deep learning-based multimodal fake user detection framework by analyzing user behavior datasets within a time retrospective window. It integrates a pre-trained Domain Large Model to represent user behavior data across multiple modalities, thereby constructing a high-generalization implicit behavior feature spectrum for users. In response to the tendency of existing fake user behavior mining to neglect time-series features, this study introduces an improved network called Sequence Interval Detection Net (SIDN) based on Sequence to Sequence (seq2seq) to characterize time interval sequence behaviors, achieving strong expressive capabilities for detecting fake behaviors within the time window. Ultimately, the amalgamation of latent behavioral features and explicit characteristics serves as the input for spectral clustering in detecting fraudulent users. The experimental results on Weibo real dataset demonstrate that the proposed model outperforms the detection utilizing explicit user features, with an improvement of 27.0% in detection accuracy.

To tackle the intricate challenges associated with the low detection accuracy of images taken by unmanned aerial vehicles (UAVs), arising from the diverse sizes and types of objects coupled with limited feature information, we present the SRE-YOLOv8 as an advanced method. Our method enhances the YOLOv8 object detection algorithm by leveraging the Swin Transformer and a lightweight residual feature pyramid network (RE-FPN) structure. Firstly, we introduce an optimized Swin Transformer module into the backbone network to preserve ample global contextual information during feature extraction and to extract a broader spectrum of features using self-attention mechanisms. Subsequently, we integrate a Residual Feature Augmentation (RFA) module and a lightweight attention mechanism named ECA, thereby transforming the original FPN structure to RE-FPN, intensifying the network’s emphasis on critical features. Additionally, an SOD (small object detection) layer is incorporated to enhance the network’s ability to recognize the spatial information of the model, thus augmenting accuracy in detecting small objects. Finally, we employ a Dynamic Head equipped with multiple attention mechanisms in the object detection head to enhance its performance in identifying low-resolution targets amidst complex backgrounds. Experimental evaluation conducted on the VisDrone2021 dataset reveals a significant advancement, showcasing an impressive 9.2% enhancement over the original YOLOv8 algorithm.

Ransomware is a type of malicious software that encrypts or locks user files and demands a high ransom. It has become a major threat to cyberspace security, especially as it continues to be developed and updated at exponential rates. Ransomware detection technology has become a focus of research on information security risk detection methods. However, current ransomware detection techniques have high false positive and false negative rates, and traditional methods ignore global word co-occurrence and correlation information between key node steps in the entire process. This poses a significant challenge for accurately identifying and detecting ransomware. We propose a ransomware detection model based on co-occurrence information adaptive diffusion learning using a Text Graph Convolutional Network (ADC-TextGCN). Specifically, ADC-TextGCN first assign self-weights to word nodes based on sensitive API call functions and preserve co-occurrence information using Point Mutual Information Theory (COIR-PMI); then our model automatically learn the optimal neighborhood through an Adaptive Diffusion Convolution (ADC) strategy, thereby improving the ability to aggregate long-distance node information across layers and enhancing the network’s ability to represent ransomware behavior. Experimental results show that our method achieves an accuracy of over 96.6% in ransomware detection, proving its effectiveness and superiority compared to traditional methods based on CNN and RNN in ransomware detection.

In the original article, we have identified an inappropriate reference in the citation section of the context. To ensure the academic rigor and accuracy of the article, we decided to replace a reference with the new one. Here is the new version of the article.

This paper tackle the challenges associated with low recognition accuracy and the detection of occlusions when identifying long-range and diminutive targets (such as UAVs). We introduce a sophisticated detection framework named UAV-YOLOv5, which amalgamates the strengths of Swin Transformer V2 and YOLOv5. Firstly, we introduce Focal-EIOU, a refinement of the K-means algorithm tailored to generate anchor boxes better suited for the current dataset, thereby improving detection performance. Second, the convolutional and pooling layers in the network with step size greater than 1 are replaced to prevent information loss during feature extraction. Then, the Swin Transformer V2 module is introduced in the Neck to improve the accuracy of the model, and the BiFormer module is introduced to improve the ability of the model to acquire global and local feature information at the same time. In addition, BiFPN is introduced to replace the original FPN structure so that the network can acquire richer semantic information and fuse features across scales more effectively. Lastly, a small target detection head is appended to the existing architecture, augmenting the model’s proficiency in detecting smaller targets with heightened precision. Furthermore, various experiments are conducted on the comprehensive dataset to verify the effectiveness of UAV-YOLOv5, achieving an average accuracy of 87

To address the challenge of balancing privacy protection with regulatory oversight in blockchain transactions, we propose a regulatable privacy protection scheme for blockchain transactions. Our scheme utilizes probabilistic public-key encryption to obscure the true identities of blockchain transaction participants. By integrating commitment schemes and zero-knowledge proof techniques with deep learning graph neural network technology, it provides privacy protection and regulatory analysis of blockchain transaction data. This approach not only prevents the leakage of sensitive transaction information, but also achieves regulatory capabilities at both macro and micro levels, ensuring the verification of the legality of transactions. By adopting an identity-based encryption system, regulatory bodies can conduct personalized supervision of blockchain transactions without storing users’ actual identities and key data, significantly reducing storage computation and key management burdens. Our scheme is independent of any particular consensus mechanism and can be applied to current blockchain technologies. Simulation experiments and complexity analysis demonstrate the practicality of the scheme.

In the past decade, deep learning has greatly increased the complexity of industrial production intelligence by virtue of its powerful learning capability. At the same time, it has also brought security challenges to the field of industrial production information networks, mainly in two aspects: production safety and network information security. The former is mainly focused on ensuring the safety of personnel behavior in the production environment, including two different categories: detection of dangerous targets and identification of dangerous behaviors. The latter focuses on the safety of industrial information systems, especially networks. In recent years, deep learning-based detection techniques have made great strides in addressing these dual problems. Therefore, this paper presents an exhaustive study on the development of deep learning-based detection methods for industrial production safety analysis and information network security problem detection. The paper presents a comprehensive taxonomy for classifying production environments and production network information, classifying and clustering prevalent industrial security challenges, with a special emphasis on the role of deep learning in insecure behavior identification and information security risk detection.We provides an in-depth analysis of the advantages, limitations, and suitable application scenarios of these two approaches. In addition, the paper provides insights into contemporary challenges and future trends in this field and concludes with a discussion of prospects for future research.

Investigating the extinction properties in dense molecular clouds is of significant importance for understanding the behavior of interstellar dust and its impact on observations. In this study, we comprehensively examined the extinction law in the Ophiuchus cloud across a wavelength range from 0.8 μm to 8 μm. To achieve this, we analyzed NIR and MIR data obtained from the UKIDSS GCS and the Spitzer c2d survey, respectively. By fitting a series of color–color diagrams, we determined color-excess ratios EJ−λ/EJ−K for seven passbands. These ratios were then directly converted to derive the relative extinction law Aλ/AK. Our findings demonstrate that the Ophiuchus cloud exhibits a characteristic of flat MIR extinction, consistent with previous studies. Additionally, our results reveal variations in the extinction law with extinction depth, indicating a flatter trend from the NIR to MIR bands as extinction increases. Notably, our analysis reveals no significant difference in the MIR extinction law among the four dark clouds: L1712, L1689, L1709, and L1688. However, distinct variations were observed in the extinction law for regions outside the dark clouds, specifically L1688N and L1688W. These regions displayed lower color-excess ratios EJ−λ/EJ−K in the Spitzer/IRAC bands. This observation lends support to the dust growth occurring in the dense regions of the Ophiuchus cloud.

In order to enhance the storage efficiency of drug traceability code information on the blockchain and improve the extraction capability of drug traceability codes on drug packaging, a detection algorithm based on an enhanced version of YOLOv5 is proposed for the drug production and transportation scenario. The proposed algorithm introduces the SPD-Conv module into the backbone network, thereby enhancing the network’s ability to extract detailed feature information. Additionally, the CA attention mechanism is incorporated into the Neck of the network, providing the network with superior feature fusion capabilities. Furthermore, the activation function in the network is replaced with the LeakyReLU activation function, reducing computational requirements during training and inference. This replacement improves the model’s test accuracy and detection speed. Experimental evaluation, conducted on a self-built dataset, demonstrates the effectiveness of the improved YOLOv5 model. The results indicate a compression of model parameters from 6.14M to 2.71M, an increase in mAP@.5 from 82.4% to 93.7%, and a boost in detection speed from 32FPS to 51FPS. These findings establish the superior performance of the enhanced model compared to the original version.

药品包装生产数据质量检测主要以人工目检与模式识别方法为主,这种方式对于缺陷特征复杂多变的判别效率低精度差.针对上述问题,在传统视觉算法基础上,为支持高精度复杂缺陷的准确分类和定位,提升工业包装信息智能处理能力,提出了基于改进深度学习网络的"识码定检缺"智能检测方法,首先利用传统模式识别对二维码进行基础分割定位,根据定位信息对药袋常见缺陷位置进行简易缺陷测算;其次采用增加通道注意力机制的目标检测改进YOLOv5算法,实现对测算位置内复杂疑难缺陷数据实时分类和定位.实验结果表明,方法在疑难复杂缺陷综合识别精度上均有提升,精确率、召回率和F1值分别是0.965、0.99和0.98,满足智能制造背景下物流包装线上生产数据精准检测和过滤,有效提升了工业异常缺陷的在线监测和质量管理水平.

The objective of this study is to propose a new regulable blockchain trusted management identity system, which can solve the fundamental problem of blockchain node identity management with large gaps in application requirements, trust models, and regulatory compliance. To achieve this goal, firstly, a hierarchical CA system framework is established, using cross-domain nodes to store hash lists of certificate chains, and achieving node identity authentication and authority management through the division of certificate functions. Secondly, the certificate designed in this study is used for cross-domain authentication and interaction of business terminals. This study designs dynamic authority access of terminals and authority-based authentication mechanism to realize identity access and cross-domain authentication of terminal entities between domains. While ensuring security and certificate validity, it reduces the number of hash operations and verification signatures and improves the efficiency of cross-domain authentication compared with existing blockchain cross-domain authentication schemes. It provides a systematic solution for identity security and privacy protection when various facilities are connected to the blockchain system.