摘要:The advancement and implementation of digital intelligence technology provide additional impetus for the transformation of educational assessment. We examine the value positioning of the comprehensive quality assessment of college students facilitated by digital intelligence from three perspectives: the varied participation of evaluation subjects, the integration of perceptions during the evaluation process, and the precise analysis of evaluation outcomes. Utilizing ternary space theory, we develop a comprehensive quality assessment model for college students grounded in the 'human-machine-thing-object-field' framework. Based on this framework, this study investigates the implementation framework for a comprehensive quality evaluation of college students facilitatted by digital intelligence. It delineates the architecture of this evaluation across five layers: physical, data, analysis, service, and interaction. Furthermore, it examines the implementation trajectory through initial, mid-term, and long-term stages, aiming to offer insights for the practical application of digital intelligence in assessing college students' comprehensive quality.
WANG Dongqi, YANG Hongshan, HUANG Bo, HE Chuanpeng, GAO Zhirong, LIU Jin
DOI:10.14188/j.1671-8836.2024.0028
摘要:Current deep learning-based methods for text sentiment classification have not fully exploited character-level information and sentiment source information, leaving room for improvement. To address this, we propose a novel method, Character-BERT-Sentiment-Attention (CBSA), which leverages the pre-trained BERT model and focuses on multi-sentiment character-level information. This approach enhances the semantic representation of text by capturing character-level features, and concatenates BERT’s pre-trained word embeddings to form a unified semantic vector. Subsequently, contextual words are integrated with four types of sentiment source information (sentiment words, negation words, degree adverbs, and conjunctions) into a gated recurrent unit(GRU) neural network. An attention mechanism is employed to generate a sentence representation that incorporates sentiment-related resources, followed by a softmax classification layer to predict sentiment polarity. By extracting both character-level and multi-sentiment source information, the proposed method captures sentiment-relevant features from diverse representational subspaces, thereby improving the accuracy of sentiment prediction. Experimental results on five benchmark datasets demonstrate that the proposed method significantly improves classification accuracy.
摘要:Most logistics parks (LPs) have low digital maturity and lack digital threads, which prevents the formation of daily collaborative operations. A general-purpose architecture and collaborative strategy for digital twin systems in LPs are proposed from the perspective of combining technology and management strategies. Firstly, the visualization requirements and mechanisms for logistics park operations are analyzed, and a five dimensional model of digital twin in logistics parks is constructed from the dimensions of digital twin system composition. Then, a universal architecture for collaborative systems between logistics parks is formed based on user dimensions, as well as the digital thread for the collaboration between LPs is designed; Secondly, from the point of implementing collaborative strategies between LPs, Evolutionary Game analysis of collaborative strategies between LPs with Cloud platforms as digital thread is conducted in a competitive and cooperative market environment. The research results show that the maturity of digital twins in LPs and the incentive degree of digital thread (Cloud platform) are directly proportional to the probability of LPs choosing collaborative strategies, and have a positive relationship with the profits of LPs. Which confirms the feasibility and positive role of digital twins in technology and operational management in the daily collaboration between LPs. The research conclusion provides reference for the collaborative operation and management of LPs.
关键词:digital twin;logistics park;collaboration strategy;cloud platform;evolutionary game
PAN Kewen, LI Na, LIU Yu, DU Ruiying, HE Kun, CHEN Jing
DOI:10.14188/j.1671-8836.2024.0163
摘要:Asynchronous Byzantine fault tolerance protocols, known for their fast and reliable advantages, are widely used to achieve agreement on transactions in unstable and adversarial blockchain networks. The state-of-the-art solutions mainly adopt the asynchronous common subset approach, which achieves agreement through two phases: dispersion and consensus. However, these solutions often suffer from high communication overhead and lack formal design for aggregating common subsets. To address these problems, this paper proposes a highly parallelized asynchronous Byzantine fault tolerance protocol featuring a three-stage asynchronous common subset framework, including dispersion, agreement, and reconstruction stages, to enhance throughput and reduce latency. In this framework, the communication overhead is effectively reduced by introducing erasure codes in the dispersion stage and using vector structure in the agreement stage. In addition, this paper proposes a common subset aggregation mechanism for efficiently aggregating subsets that have reached agreement. Experimental results show that, compared with other asynchronous Byzantine fault tolerance protocols, our approach improves transaction throughput by 1.2-3.5 times and reduces transaction latency by 13%-42%.
关键词:asynchronous agreement protocol;Byzantine fault tolerance;asynchronous common subset;blockchain
YANG Rui, SUN Mengting, QI Yuyang, WU Jiliang, MENG Qingfei, FANG Yongjin, CAO Yuliang
DOI:10.14188/j.1671-8836.2024.0185
摘要:In the quest for high-energy lithium primary batteries, Cr8O21 has emerged as a promising cathode material owing to its superior theoretical capacity. A key area of research is focused on improving the electrochemical performance of chromium oxide materials. This study presents the synthesis of Cr8O21/V2O5 composite materials via a high-temperature pyrolysis technique. Notably, the optimal electrochemical performance was achieved with the Cr8O21@15%V2O5-300 composite, derived from a CrO3 precursor with a 15% V2O5 addition, pyrolyzed at 300 ℃. This material demonstrated a remarkable discharge capacity of 389.17 mAh/g at a 0.1 C rate, representing a 10.7% enhancement over pristine Cr8O21. Electrochemical impedance spectroscopy revealed that the integration of V2O5 significantly lowered the charge transfer impedance, thereby enhancing the conductivity of the composite. Comprehensive material characterization indicated that while V2O5 did not alter the crystal structure of Cr8O21, it did induce morphological changes, leading to smaller and more uniformly distributed particles, which in turn significantly boosted the electrochemical performance of the material.
摘要:With the advancement of digital transformation, Generative Artificial Intelligence (Gen AI) has provided new technological pathways for the innovation of educational assessment paradigms. This paper delves into the logical mechanisms, application scenarios, challenges, and strategies for Gen AI-empowered comprehensive quality evaluation. Through the analysis of relevant cases, the application scenarios are categorized into multidimensional integration of evaluation content, dynamic adaptation of evaluation results, innovative integration of evaluation methods, and collaborative participation of evaluation stakeholders. Meanwhile, the paper highlights potential challenges such as deviation from evaluation objectives, insufficient capabilities of evaluation stakeholders, biased evaluation content, and ethical and privacy risks. To address these issues, strategies including value realignment, stakeholder capacity building, content refinement, and security assurance are proposed. These strategies are aimed at promoting the deep integration of Gen AI technology with comprehensive quality evaluation, driving the continuous optimization and innovative development of the educational assessment system, and supporting the cultivation of high-quality talent in the new era.
摘要:To improve the accuracy of estimating the state of health (SOH) of lithium-ion batteries and overcome the limitations of existing methods to fully characterize the details of battery decay, the paper proposes a method integrated by distance intersection over union loss (DIoUloss), simple, parameter-free attention module (SimAM) and convolutional neural network-bidirectional long short term memory network(CNN-BiLSTM) for battery SOH estimation. The incremental energy area (IEA) and charging time (T) of lithium-ion batteries are composed of IEA-T features for the battery SOH estimation. The DIoUloss loss function and SimAM mechanism are integrated into the CNN-BiLSTM model to establish the CNN-BilSTM-SIMAM battery SOH estimation model. The cyclic aging experiments of lithium-ion batteries are tested. Compared with other methods such as GRU, SVR, CNN-LSTM and CNN-BiLSTM, the proposed method can more effectively characterize the details of battery health decline. The coefficient of determination is higher than 0.96, and the maximum root-mean-square error is less than 0.020, showing favorable accuracy and efficiency.
关键词:lithium-ion battery;state of health (SOH);convolutional neural network-bidirectional long short-term memory network (CNN-BiLSTM);distance intersection over union loss (DIoUloss) function;simple, parameter-free attention module (SimAM);incremental energy
YANG Xiuzhang, PENG Guojun, WANG Chenyang, ZHOU Yilin, LI Jiachen, WU Shuai, FU Jianming
DOI:10.14188/j.1671-8836.2024.0126
摘要:Aiming at the challenges that current methods for tracing Advanced Persistent Threat(APT) attacks ignore ATT&CK(Adversarial Tactics, Techniques, and Common Knowledge) and semantic behavior enhancement, failing to combine dynamic and static perspectives for in-depth attack traceability analysis, and were easy to evade by the packed and obfuscated APT malware. To this end, a novel traceability model for APT attacks based on dynamic and static semantic behavior enhancement (APTEye) was proposed. Firstly, an APT malware sample set was constructed and preprocessed. Secondly, static and dynamic behavioral features of the malware were extracted. Then, a behavioral feature semantic enhancement and representation algorithm was designed. Attack2Vec was used to map static API features, the attack chain, and the semantic behaviors. APISeq2Vec was employed to enhance the temporal semantic relationships of dynamic API sequences, thereby mapping low-level behavior features to high-level attack patterns. Subsequently, a dynamic-static feature alignment and behavior semantic aggregation algorithm was designed to integrate dynamic and static features of APT malware. Finally, a graph attention network-based model was constructed to trace APT organizations. Experimental results show that the proposed method can effectively track APT attacks, with a precision of 92.24%, a recall of 91.85%, and an F1-score of 92.04%, superior to state-of-the-art models. In addition, the APTEye model can effectively identify fine-grained dynamic and static API functions and attack behaviors, facilitating the mapping between behaviors and the ATT&CK framework, which supports further intent inference and blocking of APT attacks.
摘要:Iron-based sulfates are characterized by low-cost and high operating voltage, making them promising cathode material for sodium-ion batteries with great application prospects. However, the synthesis of these materials typically involves high-temperature solid-state reactions, which can be energy intensive. In this work, a series of low-crystallinity Na2SO4-xFeSO4/C composite materials were prepared using a mechanical ball-milling method, and the effects of different compositions (x=1.3, 1.4, 1.5, and 2.0) on their structures and electrochemical performance were explored. The results indicated that Na2SO4-1.4FeSO4/C composite exhibited excellent rate performance and outstanding cycling stability, with an initial reversible specific capacity of 100.3 mAh/g, comparable to Na2+2xFe2-x(SO4)3 materials synthesized via high-temperature solid-state methods. After cycling 1000 cycles at a current density of 500 mA/g, the capacity retention was 93.5%. Experimental characterization and electrochemical tests showed that low crystallinity Na2SO4-1.4FeSO4/C exhibited high reaction kinetics and pseudo-capacitance contribution rate. This study presents an environmentally friendly new method for the design and synthesis of non-stoichiometric iron-based sulfate materials for sodium-ion batteries through a simple mechanical synthesis approach.
TANG Biao, ZHAO Kai, JIANG Xiaoyi, LIU Yuchen, YAN Ning
DOI:10.14188/j.1671-8836.2024.0059
摘要:The in-situ grown NiCo(PxSy)/NF bifunctional catalyst was combined with the Zn/Zn2+ redox mediator to assemble a zero-gap membraneless flow electrolyzer, enabling the alternating production of high-purity hydrogen and oxygen (above 99.5%) with stable operation for over 10 hours at a high current density of 100 mA/cm2. The Zn/Zn2+ mediator decouples the water-splitting process and offers cost advantages, while the NiCo(PxSy)/NF exhibits excellent bifunctional catalytic activity. Together, they achieve the separation of the hydrogen evolution reaction (HER) and oxygen evolution reaction (OER) in both process and time. This approach addresses the issue of hydrogen and oxygen mixing and crossover, which can lead to explosions in traditional industrial water electrolysis, providing a practical strategy for decoupling electrode reactions in water electrolysis.
关键词:water electrolysis;Zn-mediated membraneless electrolyzer;decoupling;bifunctional catalyst;hydrogen energy;sustainable energy
摘要:Adopting a macro-scale perspective of knowledge ecology, this study investigates evolutionary patterns of networked knowledge within internet-based communities. We propose a theoretical framework for knowledge population evolution, incorporating a novel Cohesion Index that synthesizes distribution density and semantic correlations of knowledge units. The methodology integrates population intensity metrics with family tree analysis. Empirical data derives from the “Consumer-Driven Educational Supply-Side Reform” module in the cMOOC course “Internet+ Education: A Dialogue Between Theory and Practice” Multidimensional analysis reveals three salient findings: 1) Knowledge evolution exhibits nonlinear cyclical progression; 2) Population development transitions through four distinct phases: Formation, Development, Stabilization and Decline; 3) Emergent phenomena include inter-population compatibility and neo-population generation. This research contributes an operational framework with quantitative metrics, advancing both theoretical understanding and practical implementation of online education ecosystems.
ZHUANG Yuan, HAN Xu, SUN Xiao, JIANG Jiale, ZHOU Jiasheng
DOI:10.14188/j.1671-8836.2024.0111
摘要:Visible light positioning (VLP) has received widespread attention in indoor positioning research and applications due to its advantages such as low cost, high accuracy, and extensive deployment in existing infrastructure. However, VLP systems experience a significant decline in localization performance when obstructed or tilted at the receiver end. To address this issue, a VLP/INS tightly coupled navigation system enhanced by adding motion constraints is proposed, which uses an inertial measurement unit for inclination estimation and applies filtering and fusion techniques at the sensor’s raw observation level to mitigate the effects of obstruction. In practical tests, the system achieved an average positioning accuracy of 10.27 cm, with accuracy improvements of 35.34% and 63.19% compared to pure VLP under occlusion and tilt situations, respectively, resulting in higher accuracy and robustness than existing methods.
XU Yaping, LI Yanyan, GUO Weitong, TANG Xiren, SHAN Yugang, LI Honghui
DOI:10.14188/j.1671-8836.2024.0164
摘要:Efficient and accurate assessment of classroom learning engagement is crucial for dynamically tracking learning progress and improving teaching quality. However, existing research faces challenges such as limited analytical dimensions and insufficient model generalizability. To address these issues, this study, leveraging deep learning technologies, proposed an intelligent assessment method for classroom learning engagement that integrates both behavioral and emotional analysis. The method encompasses the entire process, including dataset construction, deep learning model training, model evaluation, application, and statistical analysis. The results demonstrate that the trained deep learning model exhibits exceptional accuracy and robustness in both behavioral and emotional analysis tasks. Through a case study of classroom video recordings from a specific class, this research further revealed the dynamic evolution of classroom learning engagement at both individual and collective levels, validating the method's effectiveness and feasibility in real-world teaching contexts.
摘要:The navigation system relies on sensors to perceive the surrounding environment. Currently, navigation systems based on a single sensor are struggling to meet the demands of complex scenarios, leading to a shift towards multi-source sensor integration. In the process of fusing data from multiple sensors, image data processing consumes the most time and resources, significantly impacting system performance. To address the issues of single-sensor navigation systems and the time and resource consumption of image processing, a hardware control terminal for an intelligent navigation platform is designed, utilizing Global Navigation Satellite System (GNSS) Pulse Per Second (PPS) for time synchronization to achieve data fusion from multiple sensors. Additionally, an ORB (Oriented FAST and Rotated BRIEF) feature extraction accelerator is designed for the frontend of Simultaneous Localization And Mapping (SLAM) to accelerate the image processing process and improve the real-time performance of the SLAM system. Experimental results show that the hardware platform not only supports data acquisition and fusion from GNSS, Inertial Measurement Units (IMU), vision, and LiDAR but also accelerates the extraction of ORB feature points. When performing ORB feature extraction tasks, the accelerator achieves frame rates 2.7 times and 1.8 times higher than implementations on CPU and GPU platforms, respectively, while the power consumption is only 5.1% and 2.9% of them.
关键词:intelligent navigation;multi-source sensor;time synchronization;ORB(Oriented FAST and Rotated BRIEF) feature extraction;hardware accelerator
HE Hanlin, LIU Xiaofeng, YANG Jiaoyang, WANG Binlan, LI Xi, DUAN Junxin
DOI:10.14188/j.1671-8836.2024.0184
摘要:Bimetallic electrocatalysts Ag-Biy/NF with porous structure were prepared in situ by the electrodeposition with hydrogen bubble as the template. The electrocatalytic activity of Ag-Biy/NF on the debromination of bromoacetic acid was studied. When the deposition time was 6 min, the deposition voltage was 10 V, and the mass ratio of Ag+ to Bi3+ in the electrodeposition liquid was 1:1, the prepared material exhibits excellent electrocatalytic performance: It was found that the degradation rate of 25 mmol/Lbromoacetic acid at Ag-Bi1/NF-10-6 at 20 mA/cm2 was up to 100% after 4.5 hours degradation experiment, and Ag-Bi1/NF-10-6presented a stable dehalogenation performance in 8 cycles, indicating that the catalyst had an excellent electrocatalytic activity. Especially, for bromoacetic acid with a concentration of up to 133 mmol/L, Ag-Bi1/NF-10-6 also could completely remove bromine within 12 hours, and can be used in a wide range of pH, demonstrating a good industrial application prospect.
ZHANG Jianxin, HU Huijun, HU Zhenyi, LIU Maofu, ZHANG Yaofeng
DOI:10.14188/j.1671-8836.2024.0138
摘要:The goal of multimodal conversational emotion recognition is to identify the emotion of each utterance in a conversation. Existing methods for modality fusion, which are relatively simple, fail to fully capture and leverage the characteristics and information of different modalities. Furthermore, these methods primarily focus on capturing local context, and they often overlook the integration of long-range emotional information between speakers, especially in handling lengthy conversations. To address these issues, this paper proposes a Graph Neural Network Based on Multimodal Bidirectional Fusion (GMBF), which consists of a multimodal fusion module and a long-range emotion fusion module. The multimodal fusion module is composed of three bidirectional fusion submodules, each of which integrates multimodal information from both forward and reverse directions, ensuring comprehensive information fusion through progressive integration. The long-range emotion fusion module first constructs sentence-level information for the conversation, then captures long-range speaker information and incorporates it into the sentence representations, enabling the model to better understand the global emotional context. Experimental results demonstrate that the proposed method achieves superior performance on multimodal conversational emotion recognition tasks, showcasing its advantages in multimodal information fusion and global information extraction.
关键词:emotion recognition in conversation;multimodal fusion;graph neural networks
TANG Haolin, YUAN Yulin, LU Xiao, WANG Lucai, WU Chengzhong, WANG Yaonan
DOI:10.14188/j.1671-8836.2024.0143
摘要:Transformer-based models have achieved excellent performance in 3D human pose estimation tasks by encoding the connectivity between all joints to capture the global dependencies of the data. However, such methods often fail to adequately model the local dependencies of joints and suffer from the quadratic growth of computational complexity with the sequence length, leading to a waste of computational resources. To address these issues, this paper defines the human motion process as a sequence input-output process of a state-space model and proposes a 3D human pose estimation method, GMambaScanX, based on the fusion of GCN and Mamba. The GCN model captures the spatiotemporal local dependencies of human motion sequences, enhancing the model's ability to capture local relationships. The Mamba model captures the spatiotemporal long-range dependencies of human motion sequences, improving the model's ability to capture global relationships. We introduce two MambaScan scanning strategies: a sequential joint scan for temporal motion state transitions, which enhances the model's understanding of human motion features in the temporal dimension, and a spatial joint scan guided by human structural prior knowledge, which enhances the model's understanding of human structural features in the spatial dimension. GMambaScanX is trained and validated on the Human3.6M dataset, achieving an average joint position error of 39.8 mm with only 12.4% (5.3M) of the parameters of MotionBERT (SOTA). Compared to fully connected models, GMambaScanX better extracts human structural features, utilizes parameters more efficiently, and improves the model's effectiveness and efficiency.
关键词:3D human pose estimation;state-space model;graph convolutional network;human joint structure scanning
摘要:To address the limitations of inflexible coefficients in low-dimensional regular matrix encryption algorithms and the challenges associated with constructing high-dimensional encryption matrices, this paper proposes a high-dimensional generalized Arnold transform for the quantum image scrambling algorithm based on integer matrices obtained from proportional sequences. Initially, two high-dimensional integer matrices with unit determinants are constructed, and through conventional matrix multiplication techniques, a high-dimensional generalized Arnold transform matrix is derived. Subsequently, employing the universal color quantum image representation framework, this transform matrix is seamlessly integrated into the quantum image encryption process. Additionally, the inverse of the high-dimensional generalized Arnold transformation matrix is formulated and utilized in the image decryption procedure based on the universal color quantum image representation. The proposed algorithm boasts a diverse range of transformation formulas, enabling the generation of high-dimensional encryption matrices. The feasibility of this approach is exemplified through the encryption of 24-bit true-color images. Simulation results underscore the algorithm’s expansive key space, enhanced key randomness, and robust anti-attack capabilities, thereby fulfilling the stringent requirements of cryptography and demonstrating significant theoretical and practical merits.
WU Xin, YAO Wengao, ZHAN Tingting, YU Ao, WANG Zhengyang, HE Xiongying, QIN Zhiyi, LI Xi, DUAN Junxin
DOI:10.14188/j.1671-8836.2024.0161
摘要:PANI/PC@CC composite flexible electrode was prepared by potentiostatic in-situ electrodeposition of polyaniline/porous carbon (PANI/PC) on the surface of carbon cloth (CC). The effects of preparation conditions on the structure and capacitive performance of PANI/PC@CC were discussed. The results showed that porous carbon could effectively improve the rate capacity and cycle performance of PANI/PC@CC which also exhibited good flexibility. PANI/PC@CC obtained under optimal conditions had a specific capacitance up to 3 042.8 mF/cm2 at 1 mA/cm2 as well as a rate capacity up to 88.0% at 20 mA/cm2, and the capacitance retention was as high as 97.6% after 2 000 charge/discharge cycles at 15 mA/cm2, demonstrating an excellent capacitive property. Finally, PANI/PC@CC-based flexible symmetric supercapacitor also demonstrated an excellent electrochemical performance and flexibility.