This PDF file contains the front matter associated with SPIE Proceedings Volume 13289, including the Title Page, Copyright information, Table of Contents, and Committee Page.

Optical fiber communication system plays an important role in the field of data communication which has strict requirements for high-speed, reliable and real-time data transmission. Therefore, in order to ensure the security of data transmission, we should strengthen the security requirements of optical fiber communication system in this aspect. Based on the discrete logarithm problem, this paper proposes the first log-sized traceable ring signature scheme with DualRing, a new construction paradigm designed by Yuan et al. The scheme has the function of digital signature technology to ensure that the information is not tampered with, which can ensure the anonymity of real users and avoid the excessive anonymity of dishonest users. Next, we prove the security of two schemes: unforgeability, traceability and anonymity. Finally, the communication and calculation costs are analyzed, and the comparison results show that the two schemes are more efficient.

With the development of technologies such as all-optical networks, fiber-to-the-home, and automatic switching networks, optical fiber communication is experiencing a new peak. However, eavesdropping techniques targeting optical fiber communication signals are also becoming increasingly sophisticated. Additionally, in the high-speed data transmission and processing environment of all-optical networks, optical communication faces potential hazards where even brief or commonly patterned attacks could lead to significant data corruption or leakage. Therefore, the security and privacy protection of optical fiber communication cannot be overlooked. Designing an intrusion detection system capable of accurately and promptly capturing malicious traffic and taking preventive measures is crucial for maintaining the security of optical communication networks. Currently, many researchers are using machine learning and deep learning for intrusion detection. However, these methods have some drawbacks, such as manual feature selection, inability to automatically extract spatial and temporal features from traffic data, and relatively shallow network layers. To address these limitations of intrusion detection methods, we propose a model called Res-GLN. It combines Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks into a hybrid network to detect abnormal traffic in networks. It is worth noting that we designed two residual network blocks to deepen the hierarchical structure of the combined CNN and LSTM network, enabling the model to learn better feature representations without compromising performance. We conducted extensive experiments on three public network traffic datasets, and compared to other baselines, Res-GLN exhibits higher accuracy and lower false alarm rates.

Accompanied by the development of big data and cloud computing and other technologies, the traditional centralized authentication technology faces increasing security risks due to the existence of its trusted center. In order to solve the problems, blockchain technology enters the scope of research scholars. In this paper, we focus on the authentication and secure data transmission scheme of data sender to receiver via some other node in the absence of trusted center under the situation of three-party participation in the field of identity authentication. In this paper, we introduce the identity-based cryptosystem and design a three-party joint signature and authentication scheme based on the SM9 identification algorithm, combined with the decentralized, tamper-proof, publicly verifiable and other characteristics of blockchain technology. The scheme realizes the identity authentication and message integrity verification. The analysis shows that the scheme has security attributes such as decentralization and privacy protection, at the same time, we also demonstrate that the proposed scheme has the property of mitigating common network attacks.

In order to improve the security and efficiency of video encryption and improve the speed of video encryption, in this study a video encryption algorithm is proposed based on Tent chaotic mapping and Arnold mapping. Firstly, generated a chaotic sequence by using Tent mapping, and performed type conversion on this sequence; imported a video to be encrypted and read in the specified frame image, separated this image into three channels: red channel, green channel, and blue channel, combined the three channels to form a large matrix; then, performed row and column scrambling by using chaotic sequence, and then performed a secondary scrambling by using Arnold mapping; finally, performed XOR operation on the scrambled matrix to obtain the diffusion matrix, converted the diffusion matrix into three channel images, and then obtain a ciphertext frame image. The simulation results show that: The encryption algorithm is very sensitive to the initial key, and the initial value sensitivity performance reaches10^-12,the key space is 2×10²³,the average correlation coefficient of adjacent pixels in encrypted image is -0.000087, which is very close to the ideal value of 0.This encryption algorithm can well resist the attack of statistical analysis and it has a good encryption effect on video.

While metasurface technology provides a novel and more secure method for optical encryption, its constraints in channel capacity and key protection can expose it to the risk of eavesdropping. This paper proposes a novel strategy for metasurface holographic optical image obfuscated encryption based on two-dimensional discrete hyperchaotic system. Initially, a two-dimensional bi-directional feedback hyperchaotic system is constructed, which features exceptional ergodicity, sensitivity to initial conditions, and pseudo-randomness. This system is integrated with the phase structure of the metasurface to effectively generate a phase mask capable of concealing the features of the original image. Subsequently, an association matrix is formed by correlating the phase mask with the hyperchaotic sequence, and the hyperchaotic cipher stream is utilized for permutation and diffusion, significantly enhancing the security and randomness of the encrypted image. Experimental results indicate that the encryption strategy yields ciphertext images with reduced pixel correlation and increased information entropy, which can only be decrypted upon possession of the complete hyperchaotic parameters and at least 70% of the accurate phase information of metasurface.

To meet the specific demands of airborne radar and communication systems, this study introduces an integrated radar-communication waveform design based on selecting the optimal dual sequence of complementary P4-OFDM signals. This waveform leverages the cyclic shift counts of complementary P4 codes to carry communication information, achieving dual functionality in target detection and data transmission. To enhance the data transmission rate, the study innovatively developed a dual cyclic shift signal, which expands the number of states represented by P4 codes using a pair of primary and secondary original sequences. This signal offers exceptionally high speed and distance resolution for radar detection and maintains a low bit error rate in communications, making it particularly suitable for airborne environments where the priority is on communication reliability over efficiency.

The expanding academic resources require innovative search tools for efficient retrieval of relevant information. Traditional search engines often struggle to provide accurate and pertinent results for academic content. This paper proposes a specialized search engine design methodology specifically tailored to academic resources. The methodology enables the development of search engines that can effectively navigate the complexities of academic resources, aiming to enhance retrieval efficiency and accuracy.

In this article, a low-temperature cofired ceramic (LTCC) substrate integrated cavity(SIC) circularly polarized(CP) antenna array operating in 77 GHz millimeter wave range is presented. This antenna employs LTCC and substrate integrated waveguide (SIW) technologies, utilizing a SIC as the radiating element and a T-type power divider as the feeding network, forming a 2 × 2 array. By increasing the cavity aperture size and adding metal parasitic patches on the cavity's upper surface to counteract the reverse electric field, the antenna operates in the high-order TM221-mode, achieving high-gain performance. At lower frequency, the parasitic patches inclined at 45° excites induced current with phase difference of 90° through a coupling slot to create CP performance. While at higher frequency, the equal amplitude and 90° phase difference of the TM211-mode and TM121-mode within the SIC together produce circular polarization. The combination of these two mechanisms achieves wideband circularly polarized performance. The simulation results show that the proposed array antenna achieves a 27.3% 3 dB gain bandwidth within the frequency range of 65 GHz to 86 GHz, with a peak gain of 14.2 dBi. The -10 dB impedance bandwidth and 3dB axial ratio bandwidth are 20.8% (68.1 GHz to 84.1 GHz) and 21.2% (69.9 GHz to 86.2 GHz), respectively.

This study thoroughly investigates the application of computer vision technology combined with unmanned aerial vehicles (UAVs) in environmental protection management during the construction period of pumped storage power stations, aiming to enhance the level of environmental protection through real-time and efficient environmental monitoring. By constructing an integrated UAV monitoring system with advanced computer vision algorithms, this research achieves continuous monitoring of the construction area and its surrounding environment, including but not limited to vegetation coverage and water conditions. The system evaluation results demonstrate that the monitoring system meets the expected goals in terms of accuracy, real-time performance, stability, and scalability, providing reliable data support for environmental management. This not only confirms the effectiveness of integrating UAVs with computer vision technology in environmental monitoring but also provides valuable experience and data references for future applications in similar fields. This study emphasizes the potential of technology application in environmental protection, providing a new perspective and approach for the future development of environmental monitoring and management technology.

This paper delves into the extraction and secure verification technology of cable equipment information based on image processing algorithms, demonstrating how the combination of high-resolution cameras and circular LED lights effectively improves image quality to adapt to complex industrial environments. The research applies Reed-Solomon error correction algorithms and other advanced decoding techniques to ensure accurate data recovery even in the case of damaged QR codes. The article also introduces the application of digital signatures and timestamps in enhancing data anti-counterfeiting and authenticity verification, greatly enhancing the security and accuracy of the system through rigorous data formatting and verification processes. The comprehensive use of these technologies significantly enhances the efficiency, security, and reliability of QR code systems in industrial applications, which is of great significance for promoting the intelligent and automated development of equipment management.

This paper designs and applies a virtual simulation experimental teaching platform based on natural scene image recognition algorithms to enhance students' learning outcomes in the "Digital Image Processing and Applications" course. Through inquiry-based and task-driven teaching methods, the platform implements functions such as data processing, image recognition, and video processing. Evaluation data from 178 students show improvements of 70.25%, 60.90%, 65.46%, 63.85%, and 49.11% in understanding key concepts, independent learning, theory-practice integration, engineering application, and innovation ability, respectively. The results indicate that the virtual simulation experimental teaching platform significantly enhances students' comprehensive abilities, providing effective support for course teaching.

In order to enhance urban public safety and disaster response capabilities, a deep learning based urban safety monitoring system was constructed, and technical analysis and implementation of image acquisition, object detection, and data fusion were carried out. The results indicate that the system can effectively improve the identification accuracy and response speed of security threats, which is of great significance for optimizing the urban emergency management system.

Through the drone tilt photography technology for three-dimensional reconstruction of traditional villages, the basic information of the model includes geographic location, topography, spatial pattern and other information, based on which the traditional villages can be analyzed, displayed, managed and protected for research and other work. This paper takes Fanjiazhuang Village in Jinzhong City, Shanxi Province, as an example, and analyzes the traditional villages through the production of three-dimensional digital models. This study has a certain reference significance for the digital protection of traditional villages.

The utilization of an artificial intelligence (AI) algorithm based on deep learning is a valuable tool in the field of diagnostic imaging. In the past, medical images were diagnosed by medical professionals, a process that was timeconsuming and yielded limited diagnostic information. This paper proposes an automatic detection algorithm based on the YOLOv8 deep learning method to study CT images of trauma patients' arms, and its clinical effect is evaluated. The effectiveness of the proposed method was evaluated using Kaggle's fracture data, with the results confirmed by a radiologist with an accuracy of 91%. The algorithm demonstrated sufficient sensitivity to rapidly detect human upper arm fractures in a validated set of evaluated CT images.

Controlling the safety distance of crane arm lifting is a crucial requirement for high-voltage substation lifting operations. However, precise positioning is difficult to achieve with a single sensor. Therefore, this paper proposes a multi-sensor fusion approach that utilizes UWB (Ultra-Wideband) for global detection of crane and boom positions, binocular cameras to determine the distance between the top of the boom and overhead power lines, and depth point cloud cameras to detect the distance between the boom and surrounding objects in the substation environment. The integration of this information enables real-timse monitoring of operational conditions to achieve safe lifting operations. Through system testing, we achieved excellent early warning results, ensuring the safety of lifting operations and demonstrating the advanced nature of this research.

During the process of obtaining the image, the object and the camera may have a relative motion, which often results in blurred images, motion-blurred image restoration can eliminate the image blurring. In the case of the knowledge of the image degradation is already known, according to the characteristics of the blurred image caused by uniform linear motion, the relationship of the establishment between the original image and blurred image is built in this paper, which analyses degradation and recovery process of uniform fuzzy images, derives the relevant mathematical model and programs to carry out restoration based on the model, then gets the result.

In order to implement the fundamental task of cultivating people by virtue, the field of higher education is comprehensively promoting the construction of curriculum ideological and political situation. The radar equipment course covers all kinds of typical teaching equipment, which is representative in exploring the ideological and political design and practice of the course. Combined with the characteristics of radar equipment course, the organization system of radar equipment course is constructed. Distinguish between theoretical teaching and practical teaching, and design the content of ideological and political teaching respectively. Drawing lessons from the "six-step method" of ideological and political teaching, this paper focuses on the design of ideological and political teaching method in teaching activity chain. Taking the teaching content of typical radar equipment in the United States as an example, the ideological and political practice of the course is explored. After adopting the concept of course ideological and political education, the teaching effect of radar equipment course is obviously improved, which promotes the cultivation of operational high-skilled talents.

In order to facilitate the management and analysis of large-scale cyberspace news media data, mining its potential resource information. The data of 106 mainstream news media around the world was obtained from Facebook, Twitter and Reddit through web crawler technology, and the data was cleaned, processed and analyzed by library tools such as Math and Pandas to extract valuable information and trends. Using Echarts chart library and JS front-end framework to display and visualize data in multiple dimensions in the form of topological chart, line chart and pie chart, a set of big data visualization framework for news media is designed. The results show that this framework can better understand and master the dynamic situation of global mainstream news media, and provide support and reference for news media decision-making and strategic planning. Through visual presentations, users can analyze and leverage news media data from a more holistic perspective, providing important information and insights for decision makers. This visualization framework for big data for news media helps to strengthen the insight and competitiveness of the news media industry.

The traditional domain of e-commerce short video advertisements is currently grappling with the paradox of prolonged production cycles amid a surge in market demand. Simultaneously, prevalent user-friendly tools for short video ad generation face template-related issues, limiting their ability to meet enterprises' demands for personalized brand marketing. This research based on the Science of Human Affairs and the perspective of mental imagery, conducts a spatial deconstruction analysis of short video ads design processes and contents. Leveraging multimodal learning methods, the research formulates an intelligent generation design scheme for short video ads.

The objective of this study is to optimize the processing technology of Schisandra chinensis steamed with vinegar, establish HPLC fingerprint of Schisandra Chinensis before and after steaming with vinegar, and study the changes of its chemical constituents. On the basis of single factor test and orthogonal test, With 8 active substances such as schisandrin A etc as index components were used as evaluation indexes to investigate the dosage of vinegar, steaming time and dampening time to determine the best processing technology of vinegar steaming Schisandl Chinensis. Diamonsil C18 (4.6 mm×200 mm, 5 μm) column was used with acetonitrile-water as mobile phase at the flow rate of 1.0 ml·min-1. Gradient elution was performed to establish HPLC fingerprint s before and after processing. Similarity evaluation, principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were used for data processing and analysis. These results indicate that the optimal processing technology of vinegar steaming Schisandl Chinensis was 30% of material to liquid, 7 h of dampening time, 1 h of steaming time. The common peak similarity was greater than 0.93. PCA and PLS-DA showed that the chemical composition and content of raw schisan dra chinensis and vinegar steamed schisandra chinensis were different, and could be distinguished obviously, and revealed the biggest contribution of the three potential iconic chromatographic peaks. The content determination results showed that compared with raw schisandra ch inensis, the contents of schizandrol A, schizandrol B, schisantherin A, schizandrin A increased relatively after vinegar steaming, while the contents of schizandrin B decreased relatively after vinegar steaming. The changes of schisanheol and schizandrin C before and after vinegar steaming were not obvious, and 5-HMF was unique in processed products. the content of raw schisandra chinensis was little. This study lays an experimental foundation for the formulation of processing procedure and quality standard of vinegar steamed Schisandra chinensis, and provided data support for the development of health food for improving sleep function.

Artificial intelligence (AI) is a computer technology that has a human way of thinking and can imitate the human brain. The integration of AI into medical imaging has revolutionized diagnosis and treatment, providing significant advances in diagnostic accuracy, treatment planning, and patient care. In the context of widespread global cancer challenges, this study will focus on the effectiveness of AI technology for early detection in the field of imaging and histological image analysis, and review strategies for using image processing technology to detect tumors. We discussed the latest AIdriven technologies, including deep learning algorithms, computer-aided detection systems (CAD), and computer-aided diagnosis (CADx), which excel at processing large data sets to produce accurate and valid results in cancer detection, helping imaging physicians accurately identify cancer, and enhancing early cancer diagnosis. And then improve the patient's prognosis through therapeutic intervention.

Firstly, this paper deeply studies the deep learning theory and its application in the field of image processing. By constructing an efficient convolutional neural network model, it can realize the automatic extraction and recognition of the spatial position and shape of the point-selected equipment of the transmission line. On this basis, the powerful learning ability of the deep learning model is used to accurately judge the operating status of the point-selection equipment of the transmission line, effectively improving the accuracy and efficiency of hidden danger detection.The research in this paper not only enriches the application research of deep learning in the field of power system, but also provides a new idea and method for intelligent monitoring and hidden danger detection of transmission line point equipment, which has important theoretical significance and practical value.

Three-dimensional urban models have extensive applications in various fields, but efficient construction and smooth display on web platforms are critical issues. This paper proposes a method for rapid construction of massive 3D models based on deep learning, designing an end-to-end convolutional neural network that integrates remote sensing imagery and geographic information data to automatically generate large-scale building models. Meanwhile, strategies such as model optimization, data compression, progressive transmission, and GPU parallel rendering are adopted to address web loading performance issues. Experiments demonstrate that this method can efficiently construct models covering 4 million buildings across the city, meeting mapping standards, and achieve rapid loading and smooth rendering of 3D scenes on different hardware environments, providing strong support for the development in related fields.

Efficient visualization of 3D models plays a crucial role in various fields such as power grids, but it faces challenges of massive data and rendering performance. This paper proposes a technique for model optimization and visual domain loading based on deep learning to enhance the efficiency of visualizing massive power grid 3D models. We design a 3D model simplification algorithm based on an encoder-decoder network, which significantly compresses model data while preserving geometric details. Additionally, we develop a visual domain prediction network based on attention mechanisms to achieve dynamic visual domain loading. Experimental results demonstrate the outstanding performance of this technique: a 64.2% reduction in memory usage, a 57.9% increase in rendering frame rates, and a 68% frame rate improvement while maintaining 95% visual quality, providing support for the efficient transmission and rendering of massive complex 3D models. Future work will focus on expanding related algorithms, enhancing interpretability, advancing 3D data processing, and visualization technologies to facilitate industrial digital transformation.

This paper proposes a method for analyzing the impact of power outages on users and predicting losses based on deep learning. By processing heterogeneous data from multiple sources, a high-quality dataset is constructed, and algorithms such as MLP, Random Forest, and SVR are used to train models for user classification and loss prediction. Based on this, a comprehensive analysis system integrating data management, model analysis, and visualization display is designed and implemented. Through testing, the system's usability, stability, and security are verified. This research provides a datadriven analysis tool for power companies and government decision-making departments, with important theoretical and practical significance.

Power outage events can have serious impacts on users' production and life, leading to economic losses. In this paper, a method for analyzing the impact and loss of power outage users based on data mining is proposed to address this issue. Firstly, power outage event data and user data are collected and preprocessed. Then, algorithms such as association rule mining, cluster analysis, and decision trees are applied to discover patterns of power outage events, analyze the degree of user impact, and estimate potential economic losses. Empirical studies show that this method performs best in evaluating household users and needs improvement in assessing industrial and commercial users. The research results provide data support for power supply companies to formulate response measures and optimize power supply methods, thereby improving grid reliability and disaster resilience.

The application of deep learning technology in the analysis and optimization of athlete performance has garnered increasing attention. This paper reviews the research progress of deep learning in areas such as motion capture and recognition, physical fitness and health monitoring, competition data analysis, training plan generation and adjustment, movement and tactical improvement, as well as injury prevention and recovery. Through experimental studies, deep learning models have demonstrated the ability to automatically extract key features from massive sports data, generate intelligent analysis results, and provide optimization suggestions, thus offering robust technical support for athletes and coaches. Despite facing challenges in applying deep learning to practical sports scenarios, these research findings have opened up new directions for sports science research. In the future, deep learning technology is expected to continue playing an important role in the field of sports, promoting the scientific and modern development of sports.

In order to explore the application of deep learning technology in natural language processing, this article provides a detailed analysis of various algorithms, including language models, syntactic analysis, semantic understanding, and text generation. The results indicate that by adopting advanced neural network models and rich datasets, the accuracy and efficiency of language processing can be significantly improved, thereby promoting the performance and adaptability of intelligent systems in practical applications.

This paper delves into the design and analysis of output feedback controllers for dissipative linear time-varying singular systems that exhibit state-space symmetry. We detail the development of both static and dynamic symmetrical output feedback controllers, highlighting their design processes and the necessary and sufficient conditions for achieving strict dissipation. Special attention is given to the symmetry conditions that govern the effectiveness of these controllers. Additionally, we establish the criteria for system admissibility and dissipation through matrix inequalities, supported by rigorous theoretical proofs. The practical applicability of our theoretical findings is demonstrated through the implementation and testing of the controllers in simulated environments, ensuring their performance under various operational conditions. Our results indicate that the designed controllers effectively maintain the desired system properties, contributing to the stability and robust performance of time-varying singular systems.

In order to explore the application of deep learning technology in digital art creation, this study uses Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and Style Transfer techniques to analyze their ability to imitate and innovate artistic styles and their impact on cultural identity. The results indicate that these models can effectively generate images with high artistic and cultural expression, providing a new perspective for the integration of art and technology in the future.

In order to explore the application effect of neural networks in ethical decision-making in artificial intelligence, a comprehensive decision analysis system was designed and implemented. A neural network model adapted to complex ethical issues was constructed by collecting and preprocessing multiple data sources, and its decision output was evaluated in detail. The results indicate that although the model has shown good performance in handling structured and real-time data, further optimization is needed in terms of transparency and adaptability in ethical decision-making.

With the widespread application of machine learning, privacy protection of training data has become particularly important. Trusted Execution Environment (TEE) is a technical approach for privacy protection, which has the advantage of high training efficiency. However, most current TEEs are implemented in CPU environments, and training neural networks on CPUs is inefficient, leading to a search for methods to accelerate the computation of TEEs. The current schemes proposed for acceleration either execute on a single machine but rely on model parameters for precomputation in outsourcing algorithms, making it impossible to protect the privacy of neural network training, or execute in a multi-machine, multi-GPU environment, allowing for neural network training but requiring communication among multiple machines. In this paper, we propose the MOTG(Matrix Outsourcing from TEE to GPU) framework, which executes on a single machine and can complete neural network training, with clear advantages over current technologies. The main approach adopted by MOTG is an improved matrix multiplication outsourcing algorithm, where the precomputation part of the algorithm is independent of model parameters, enabling the support for dynamically changing model parameters in neural network training scenarios. In practical tests, we applied the MOTG framework to the MNIST and CIFAR-10 datasets, using a 5-layer deep learning neural network and the classic VGG-11 neural network, respectively, without any precomputation, achieving speedup ratios of 1.27 times and 7.19 times, respectively.

The injection of false data attack (FDIA) poses a serious threat to smart grids, and accurate detection of FDIAs is crucial for the secure and stable operation of the grid. A promising approach for FDIA detection based on edge computing is to adopt a scheme that involves prediction before classification, taking into account the temporal correlations of measurement data. This paper proposes a FDIA detection model based on temporal convolutional network (TCN). The model effectively captures long-term dependencies in sequential data, thereby enhancing data prediction performance. Simulation results demonstrate that TCN exhibits superior detection performance under common neural network architectures, compared to typical recurrent neural network (RNN) models. This improvement facilitates highlighting the differences between injected false data and genuine data during the subsequent classification stage, making anomaly detection algorithms more capable of identifying abnormal situations and thereby enhancing the accuracy of FDIA detection.

Lung cancer is the leading cause of all cancer deaths. Assessment of histopathological images by a pathologist is the gold standard for lung cancer diagnosis. However, the number of qualified pathologists is too small to meet the substantial clinical demands. This study aimed to develop an automated lung cancer detection framework using while-slide histopathology images. The algorithm development consisted of the data splitting, data preprocessing, deep learning models development, training and inference processes. Two different U-Net variants (U-Net and U-Net++) with two different encoders (ResNet34 and DenseNet121) were selected as base models, and two loss functions including dice loss and weighted binary cross entropy loss were used during training. Unweighted average was used to combine results of multiple base models. On the test dataset, the ensemble model using 5X magnification and 512X512 patches obtained an accuracy, sensitivity, specificity and dice similarity coefficient of 0.934, 0.877, 0.948, 0.840, respectively. Except for the specificity of 10X magnification being slightly higher than that of 5X magnification, no matter what model type, encoder, loss function and performance metric were used, the performances of using the 5X magnification outperformed those of using the 10x and 20x magnifications. This algorithm achieved satisfactory results. Moreover, extensive experiments indicated that using 5X magnification 512X512 patches is a good choice in automated lung cancer detection. In the future, after improving the generalizability of this framework in real clinical settings, this framework can be used to assist histologists in their daily work.

In order to improve the security and efficiency of computer network communication, this article analyzes the application of deep learning technology in network communication fault detection and processing. By introducing convolutional neural networks and long short-term memory networks, research focuses on network fault prediction, feature extraction, and security reinforcement. The results indicate that combining these technologies can significantly improve the stability and security protection capabilities of the network, providing effective improvement strategies for network communication technology.

This research employed feature engineering techniques to preprocess an original stock dataset, followed by the introduction of a decision tree prediction model for forecasting the dataset. Experimental results demonstrate an enhancement in predictive performance, offering a more effective analytical tool for forecasting stock market trends. This approach also serves as an inspiration in fields such as optoelectronic signal processing, optical image recognition, and optical computation and processing.