卷 7 ， 期 3 ， 2024
- 摘要：The latest advances in deep neural networks demonstrate their ability to learn visual models of large-scale datasets. However, there is currently no publicly available dataset for road damage detection in China. The data used in this experiment is captured by a bicycle recorder. The image resolution is severely affected by vehicle occlusion, lighting changes, vehicle shaking, and other factors, resulting in a huge amount of data. However, manually selecting qualified data is time-consuming and laborious. This article proposes a new detection method that uses the feature transfer network (FTN) to reconstruct road damage images, sharpens image features while using a small amount of data, and then uses the Faster Regions with CNN features (Faster RCNN) network for damage detection. The experimental results show that using feature style transfer to reconstruct images for damage detection improves the average accuracy of different road damage category instances by 2%,demonstrating the effectiveness of the framework.关键词：road damage;object detection;feature style transfer;Faster-RCNN187|96|0更新时间：2025-04-08
- 摘要：In the era of information explosion, effectively managing and utilizing vast amounts of knowledge has become a critical issue. One of the core tasks of knowledge management systems is knowledge utilization, with search being the primary means. The effectiveness of search ranking directly impacts the efficiency of knowledge utilization. Current learning to ranking algorithms have significantly improved rankings by incorporating behavioral, semantic, and business features. However, the reasons behind the given search result rankings are often a black box to users, negatively affecting their experience. To address this issue, this paper proposes an explainable information combined deep learning to rank algorithm (EI-DLTR). This method leverages the powerful learning capabilities of deep learning to jointly model the reasons behind knowledge-related rankings. It provides users with well-ranked knowledge results while offering fact-based explanations for the rankings, thereby improving both knowledge ranking effectiveness and user experience. This algorithm is the first to integrate deep learning to rank with ranking reason and apply it in the field of knowledge search. Significant improvements are achieved in comparisons with similar algorithms that do not consider ranking reasons, as well as in online A/B tests.关键词：knowledge search;learning to rank;explainable ranking;Deep learning2|0|0更新时间：2025-04-08
- 摘要：This study focuses on the automatic identification of power theft and leakage users, and proposes an innovative algorithm. By using deep feature mining techniques, key features such as electricity trend, line loss, and alarm types are extracted from power data to construct a multidimensional feature system. Using CNN (Convolutional Neural Networks) to learn features and establish a theft and leakage user recognition model. For the theft of electricity in front of the meter, non parametric statistical methods are used to analyze the distribution of characteristic values of historical and monitoring data, and dynamically determine the warning threshold. Verified by the real user dataset of State Grid Corporation of China, the algorithm has an accuracy rate of over 90% and a recall rate of over 85%. It has high accuracy and practicality, and is suitable for various power data monitoring systems, such as wireless meter reading systems. It can effectively improve the anti theft ability of power enterprises, maintain power supply order and grid safety.关键词：convolutional neural network;stealing electric leakage from users;automatic recognition;electricity consumption data;pattern recognition2|0|0更新时间：2025-04-08
- 摘要：In order to solve the problem of data imbalance in stroke risk prediction, the purpose of this study is to improve the model's ability to identify minority classes by data expansion method. SMOTE is used to expand the training data, and the experiment is carried out with logistic regression, support vector machine, decision tree, K-nearest neighbor, random forest, gradient elevator, XGBoost and other machine learning models. The effect of SMOTE technique on the predictive power of the model is evaluated by comparing the performance of the model on the original unbalanced data set and the SMOTE treated data set.The experimental results show that on the original unbalanced data set, the model is generally difficult to identify a few classes. In the dataset treated by SMOTE, the accuracy, G-mean and F1 values of each model have improved significantly. In particular, tree-based models and integration methods show higher effectiveness in dealing with unbalanced data.关键词：Imbalanced datasets;SMOTE;Ensemble learning;Non-ensemble learning;Disease prevention59|176|0更新时间：2025-04-08
- 摘要：This paper proposes a hierarchical management method of DNS (Domain Name System) blocked domain names based on big data analysis. This method collects logs through the traffic mirror technology, which realizes real-time and efficient collection and analysis of DNS log data, avoids network congestion and inefficient analysis, and provides a strong guarantee for the security and stability of the Internet ecosystem. Compared with traditional DNS blocking strategies, this method not only reduces computational costs and resolution latency, but also improves system stability and security. The number of effective domain name blocking measures has decreased by 97.5%, resolution latency has decreased by 98%, and memory usage and CPU (Central Processing Unit) utilization have also decreased by 5%. The innovation lies in the use of traffic mirroring technology for log collection, which enables real-time and efficient collection and analysis of DNS log data, By introducing machine learning algorithms and real-time data analysis techniques, intelligent classification and popularity evaluation of malicious domain names have been achieved, enabling precise management and optimization of domain name blocking. A hierarchical management strategy has been proposed, which formulates corresponding deletion rules based on the classification results and popularity scores of domain names, achieving refined management of blocked domain names.关键词：DNS security;block domain names;big data analysis;refined management;heat management;machine learning2|0|0更新时间：2025-04-08
RESEARCH PAPER
- 摘要：Due to its advantages of flexibility, scalability and independence, the microservice architecture plays an important role in the integration of IT resources after mergers and acquisitions. By splitting large applications into small, independent service units, this architecture not only improves system maintainability and scalability, but also enhances system deployment flexibility.In the process of merger and acquisition (M&A), the IT systems of different companies are often different.Microservice architecture can realize rapid deployment and flexible management through mechanisms such as service discovery, registration and configuration management, so as to effectively integrate these heterogeneous resources. This paper examines the function of a unified authentication API (Application Program Interface) gateway in microservices architecture, specifically focusing on the design of a unified authentication system by using Zuul and OAuth2. The unified authentication API (Application Program Interface) gateway provides centralized authentication management, simplifies interactions between clients and microservices, enhances system security and consistency, reduces IT integration costs, and significantly improves IT operational efficiency following a business M&A.关键词：IT integration;microservices architecture;API gateway;Zuul;unified authentication;merger and acquisition （M&A）2|0|0更新时间：2025-04-08
- 摘要：In order to solve the technical difficulties such as relatively large volume and poor EMC (ElectroMagnetic Compatibility) of MEMS (Micro-Electro-Mechanical System) microphone chips, as well as the parasitic capacitance and inductance introduced by chip connections, soldering points, etc., research is conducted on flat capacitance materials and their preparation methods. The flat capacitance material FCM1121 prepared in this study, compared with mainstream products in the field of flat capacitance materials, endows MEMS microphone chips with equally excellent basic performance, as well as better RF (Radio Frequency) anti-interference and other performance advantages. It also provides good reference opinions for the research and development of flat capacitance materials in other fields.关键词：MEMS microphone;chips;RF anti-interference;flat capacitance material243|141|0更新时间：2025-04-08
CORRESPONDENCE
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09064830号-19
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
- Total Visits：71687 Visits Today：118
0