Digital Commons Network^™

Addressing The Problems Of Data Variations, Quality, And Scarcity In Training Deep Neural Networks, Jian Sun

Electronic Theses and Dissertations

The performance of deep neural networks (DNNs) is strongly influenced by the characteristics and quality of the underlying datasets. This Ph.D. dissertation addresses three pervasive data challenges-imbalance, quality degradation, and scarcity-that commonly hinder the effectiveness of DNNs in computer vision (CV) and natural language processing (NLP) applications.

Class imbalance remains one of the most frequent causes of degraded model generalization. While Focal Loss effectively mitigates inter-class imbalance by assigning higher weights to minority classes, it struggles with intra-class imbalance, particularly in video datasets where longer clips dominate feature representation. To address this, I implement and utilize …

Secureai: Toward Experiential Security And Privacy Training For Ai Practitioners, Mujtaba Nazari, Eric Chan-Tin, Loretta Stalans, Mohammed Abuhamad

Computer Science: Faculty Publications and Other Works

The rapid adoption of artificial intelligence across industries has outpaced security and privacy training for AI practitioners. This paper presents methods, modules, and findings from an experiential training program designed to address security and privacy challenges in AI systems development and deployment. We conducted two program iterations: a comprehensive 12-workshop series (May-October 2024) and a condensed 6-workshop format (January-February 2025). The program combined expert-led panel sessions with hands-on laboratory activities, engaging 78 participants from diverse professional backgrounds. Evaluation through pre- and post-evaluation surveys and qualitative observations revealed improvements in cybersecurity knowledge and AI security awareness. Participants demonstrated enhanced ability to …

Analogy2kg: An Automatic Pipeline For Deriving Knowledge Graphs From Long-Text Analogies, Kara Combs, Lance E. Champagne, Bruce A. Cox, Christine M. Schubert Kabban, Trevor Bihl, Grace Lemming

Faculty Publications

Analogical reasoning is an increasingly popular, lightweight solution to enable large language model (LLM)-level reasoning without computational complexity. Still, it has yet to be adopted due to its reliance on strictly hand-formatted data. Therefore, we propose Analogy2KG (“Analogy to Knowledge Graph”), as an automatic pipeline that transforms text into a KG format via a fine-tuned version of information extraction (IE) algorithms for long-text analogies. The need to verify that the complex underlying analogical structure of the data is maintained was done via paired samples tests in the creation and validation of this pipeline. Graph density was used to evaluate the …

Deep Learning Based Routing And Clustering Approaches For Energy Optimization In 5g Wireless Sensor Networks: A Review, Ghufraan Ali Mohammad Jawad, Ahmed M. Al-Salih

Journal of Intelligent Informatics, Networking, and Cybersecurity

Next-generation Wireless Sensor Networks (WSNs), including 5G and Beyond-5G networks, have made significant progress. However, their development requires re-evaluating intelligent communication approaches to meet the growing demands for higher data transmission rates, more efficient spectral utilization, and reduced energy consumption. Scalability issues related to energy efficiency remain a critical concern for WSNs, which remain integral to the digital revolution. This paper investigates the role of Deep Learning (DL) methods in improving energy efficiency for routing and clustering in 5G WSNs. It is pertinent to note that Deep Q-Networks (DQNs) and their variants have significantly enhanced Cluster Heads (CHs) selection schemes …

Clinical Prediction Of Posttreatment Migraine Recurrence Using Biofeedback Data: A Machine Learning Framework For Enhanced Patient Stratification And Treatment Monitoring, Shibbir Ahmed Arif, Ferdib-Al-Islam, Mehidy Hasan Sium

School of Computing Faculty Scholarship and Creative Works

Migraine is a complex neurological disorder with significant implications for individual well-being and public health. Predicting migraine occurrences after treatment is crucial for evaluating therapeutic efficacy and enabling personalized care, yet remains largely underexplored. This study proposes a robust machine learning framework to predict posttreatment migraine headache occurrences using real-world headache log data collected from 133 patients undergoing biofeedback therapy. The methodology includes rigorous data preprocessing, outlier removal via the interquartile range (IQR) method, and class imbalance correction through the synthetic minority oversampling technique (SMOTE). A total of 10 classical and a hybrid ensemble machine learning models were developed and …

Recovery And Validation Of Fragmented Rar Files Through The Scalpel3 Application, Thomas L. Landaiche Iii

LSU Master's Theses

A core component of filesystems includes an address table or other means of tracking metadata on where a given file resides within a disk image. This is crucial for regular operation of a computer, or disk analysis in digital forensics. When filesystem information is missing or corrupted, locating saved files on a disk becomes challenging. Whether the disk was accidentally wiped or intentionally tampered with, data still present on the disk, even while untracked, can possibly be recovered beyond what the filesystem registers. File carving in digital forensics is important for these scenarios when data recovery is necessary but address …

From Network Packets To Physical Consequences: Network-Level Spoofing And Loss Of Visibility In Industrial Control Systems, Lillian B. Beck

LSU Master's Theses

Industrial Control Systems (ICS) are the foundation of critical infrastructure as they provide resources like clean water, electricity, and natural gas to support our everyday functions. Given their essential role in modern society, the failure or compromise of ICS systems can lead to significant negative consequences and public safety impacts. Understanding how these systems can be targeted and exploited by malicious actors is necessary to improve their security and resilience.

This research examines how an attacker could manipulate a real-world ICS network by targeting a fully functional Natural Gas (NG) Compressor Station Platform that emulates real operations. This research led …

Science And Technology Communication And Construction Of China’S International Discourse System: Strategic Significance, Current Challenges, And Innovative Pathways, Haonan Du, Ting Wang

Bulletin of Chinese Academy of Sciences (Chinese Version)

Against the backdrop of a new round of scientific and technological revolution with the ongoing restructuring of the global governance system, science and technology have become central to competition in national comprehensive competitiveness and struggles for international discourse power. China’s science and technology have achieved a historic leap from “catching up” to “running alongside” and even “leading in certain areas”, providing solid strength support for the construction of an international discourse system. However, the transformation of scientific and technological advantages into discourse advantages remains insufficient, and the international community holds a dual perception of China’s technological advancements—recognition alongside concerns. The …

Understanding Key Lssues In Current Artificial Intelligence Development And Governance, Meng Li

Bulletin of Chinese Academy of Sciences (Chinese Version)

At present, as global artificial intelligence advances vigorously, the industry is engaged in intense discussions about the future direction of its development and governance, where certain consensus has been reached alongside notable divergences and controversies. This study focuses on several key contentious issues in the current development and governance of artificial intelligence, analyzes their inherent logic by combining the laws of scientific and technological progress, and puts forward corresponding solutions. Regarding the potential bubble risk, it emphasizes the need to strengthen the application of artificial intelligence in the real economy to channel and guide its development momentum. For the future …

Comparative Study On Ai Talent Cultivation In China And The United States: Strategies And Recommendations, Lexuan Li, Ke Wen, Wenjie Liu, Wei Shen, Zhenguo Li

Bulletin of Chinese Academy of Sciences (Chinese Version)

Artificial intelligence (AI), as the core driving force of the new industrial revolution, has become a strategic pillar for enhancing national competitiveness, with the cultivation of AI talent being a decisive factor. This study systematically compares the AI talent cultivation systems of China and the United States from three perspectives: strategic planning, formal education, and practical domains. The findings reveal that the U.S. has continuously and systematically advanced AI talent cultivation plans at the national level, granted institutions substantial autonomy in talent cultivation with deep integration of industry-academia-research collaboration, and established a well-developed AI innovation and entrepreneurship environment. Accordingly, the …

Research And Reflections On Model Of China’S Ai Open-Source Innovation Ecosystem, Yuntao Long, Haibo Liu, Qigang Zhu, Xudong Ren, Yanjun Wu

Bulletin of Chinese Academy of Sciences (Chinese Version)

Open-source innovation, leveraging its strengths in collective collaboration and agile development, is increasingly becoming a core driver reshaping innovation paradigms and industrial ecosystems in the global artificial intelligence field. As a significant force in global technological development, China is actively promoting technology enterprises, research institutes, universities, and other diverse entities to deeply integrate into the construction of the global AI open-source innovation system through policy guidance, community building, and breakthroughs in large models. Based on the practical experience of building China’s AI open-source innovation ecosystem, this study systematically examines the current development status and challenges. By analyzing the operational models …

Transformation Of Open-Source Innovation Governance Of Large Models, Zhe Wang, Jinsong Cai

Bulletin of Chinese Academy of Sciences (Chinese Version)

Artificial general intelligence (AGI), represented by large-scale models, is a transformative driver of new-quality productive forces, making the effective construction and governance of open-source innovation commons increasingly critical. Yet traditional open-source models are under strain, and existing commons theories require further development. Model training costs rise exponentially with each iteration, while a sustainable profit cycle from innovation to industrial application has not been established, rendering long-term cost-bearing infeasible for individual organizations. The traditional community-enterprise collaboration model, once successful in generating ecosystem value, now faces escalating computing and data costs, uncertain commercial returns, and governance risks related to social and national …

Research On Signal Perception System Of Key Technology Regulation In The United States, Kaile Wang, Hao Liu, Yunwei Chen

Bulletin of Chinese Academy of Sciences (Chinese Version)

The purpose of the study is to analyze the operation mechanism of the key technology control signal perception system, and reveal the deep logic and basic procedures of the United States’ external technical containment. Through literature review and information mining, this study analyzes the structural features of the U.S. key technology control signal perception system, such as “front-end technical perception and technology locking”, “mid-end technical evaluation and risk identification”, and “back-end technical control response and dynamic sanctions”. It further explores the participating subjects and role positioning of the United States in key technology control signal perception, the sources and mechanisms …

A Taxonomy Of Generative Models With A Focus On Diffusion Models And Denoising Techniques, Aditi Singh, Nikhil Kumar Chatta, Yuvaraj Vagula, Abul Ehtesham, Saket Kumar, Tala Talaei Khoei

Computer Science Faculty Publications

Diffusion models have emerged as a powerful class of generative models, demonstrating impressive results across visual domains such as image and video synthesis. This survey provides a comprehensive taxonomy of generative models, with a particular focus on diffusion models and their applications in enhancing visual fidelity for text-to-image and text-to-video generation. We discuss the theoretical foundations of diffusion models, including their formulation through stochastic differential equations, and analyze the forward noising and reverse denoising processes that enable stable training and high-quality generation. The survey further categorizes diffusion architectures, including pixel-space and latent-space models, and examines their design choices, training strategies, …

Graph Neural Networks Predict Anti-Epileptic Treatment Response In Non-Lesional Infantile Epileptic Spasm Syndrome, Karen K. Wurzel

Master's Theses (2009 -)

Background: With an incidence rate of approximately three out of 10,000 live births, Infantile Epileptic Spasm Syndrome (IESS) is a rare form of epilepsy with about 2,000 to 2,500 new cases in the United States annually. Each year, thousands of infants and families face uncertainty, often requiring multiple medication regimens before finding an effective treatment. This thesis uses pretreatment structural brain connectivity data to train Graph Neural Networks (GNN) to predict antiepileptic drug response in infants with IESS, aiming to reduce the trial-and-error approach and improve care for those with IESS. Method: The dataset used in this thesis contains de-identified …

Deadmap: Open Yet Locked—Key-Controlled Dataset Protection With Feature-Level Collisions, Ting Yang, Mahabubur Rahman Miraj, Xinyu Lei, Nankun Mu, Hongyu Huang

Michigan Tech Publications

The performance of deep learning models relies heavily on high-quality datasets. However, under the prevailing paradigm of pretrained-model–based applications, once a dataset is publicly released, data owners largely lose control over how it is used for model training. Existing data protection approaches either focus on post hoc accountability or irreversibly destroy the training utility of the data through perturbations, making it difficult to simultaneously satisfy the dual requirements of offline public sharing, authorized and controllable usage. In this paper, we propose DeadMap, a reversible training usability control framework designed for model fine-tuning scenarios. DeadMap introduces a secret label permutation combined …

Research On Inter-Satellite Topology Design And Simulation Of Giant Leo Constellation Network With Consistent Pattern, Zhicheng Li, Shuaijun Liu, Lixiang Liu

Journal of System Simulation

Abstract: The giant low earth orbit (LEO) constellation network uses inter-satellite links to form an intersatellite topology, realizing the transmission of data between satellites. In order to adapt to the nature of uniform and symmetrical distribution of satellites in the constellation, this paper used a consistent connection pattern between satellites to construct an inter-satellite topology, and by analyzing the arrangement of non-mirror links in the constellation, it was found that the connection method of each link of the satellite itself could be independent of each other, which reduced the simulation complexity and the solution space of the inter-satellite topology. …

Design And Verification Of Manned-Unmanned Collaborative Combat Capability System Based On Mbse, Fangbo Wang, Jian Guo, Chenglie Du, Yifan Liu, Pengpeng Zhang

Journal of System Simulation

Abstract: The traditional model-based systems engineering (MBSE) method has problems of failing to fully exhibit complex combat logics in manned-unmanned collaborative combat system modeling, neglecting the scenario constraints in interface modeling, and requiring long-term and costly algorithm verification. In order to solve the problems, a methodology and design tool based on MBSE was proposed. An integrated verification method of a system's operational logic, interface design, and algorithmic design was constructed, thus providing a digital and rapidly iterative verification approach for system simulation. A verification environment for multiple key algorithm simulations was established, effectively reducing the economic cost of building verification …

Review Of 3d Human Reconstruction Methods Empowering Vr/Ar, Lisha Zhang, Yuchi Huo, Qi Ye, Anjun Chen, Shihui Guo, Jiming Chen

Journal of System Simulation

Abstract: 3D human reconstruction is critical for VR/AR. Early methods relied on multi-view cameras and depth sensors but were costly. Mid-term approaches using parametric human models enabled efficient single-image reconstruction, while implicit neural representations improved fidelity yet suffered from low efficiency. Currently, 3D Gaussian Splatting achieves high accuracy and real-time rendering as a new paradigm. Challenges include detail distortion and limited generalization, and future development will focus on VR/AR integration.

Neural Radiance Fields Based On Explicit Feature Matching And Scaled Dot-Product Attention, Mingwei Cao, Fengna Wang, Zilong Wang, Haifeng Zhao

Journal of System Simulation

Abstract: To address the problems that neural radiance fields(NeRF) are prone to artifacts and texture blurring in novel view synthesis under sparse view input and complex scenes, this paper proposed neural radiance fields based on explicit feature matching and scaled dot-product attention(EMD-NeRF). A multiscale feature extraction network was used to extract multi-scale feature information from the input sparse views. A fusion dot-product module was utilized to calculate view interaction information as a shared branch. Cosine similarity was adopted as a matching clue for similarity embedding volume rendering. A regularization loss function was used to enhance the quality of the scene …

Analytical Reentry Guidance Method Based On Lift-To-Drag Ratio-Velocity Profile, Weibo Sun, Ping Ma, Yuxuan Wang, Songyan Wang, Tao Chao

Journal of System Simulation

Abstract: An analytical reentry guidance method based on the lift-to-drag ratio and velocity profile was proposed to address the challenges of precise flight time control and prohibited area avoidance for the hypersonic glide flight vehicle during the reentry phase. A reduced-order reentry motion model was established; the Chebyshev series was used to approximate its nonlinear integral terms and differential equations; analytical expressions for the glide trajectory were yielded. The analytical relationships among flight time, remaining range, and the lift-to-drag ratio and velocity profile were derived. The optimization problem of lift-to-drag ratio and velocity profile was converted into a segmented parameter …

Research And Analysis Of Algorithm For Detecting Surface Defects On Automotive Wheel Hubs Based On Ccl-Yolov8, Yanjun Chen, Min Zhou, Meng Zha, Meizhou Zhang

Journal of System Simulation

Abstract: To address the challenges such as low detection efficiency, difficulties in identifying small defects, and poor accuracy in detecting surface defects on automotive wheel hubs, a lightweight neural network called CCL-YOLOv8 was proposed based on an improved YOLOv8n architecture. A synergistic improvement in both detection accuracy and efficiency was achieved through a three-stage model optimization strategy. A convolutional attention fusion module was introduced, which integrated convolution operations with self-attention mechanisms, thereby enhancing the model's ability to capture local features of small defects while perceiving global context under low signal-to-noise ratio conditions.A C2f-Star module was constructed to reduce computational overhead …

Simulation And Multi-Perspective Recognition Algorithm For Typical Trajectory Shapes, Xuejian Feng, Han Ding, Yiqi Tong, Chaoying Huo, Yanjin Zhang

Journal of System Simulation

Abstract: Current trajectory simulation methods inadequately address geometric shape features and kinematic properties of the target trajectory. To bridge this gap, a target trajectory shape simulation algorithm based on kinematic laws was proposed. The polar coordinate equations and curvature equations of multiple trajectories were integrated. The aircraft state parameters were solved by combining kinematic equations. Angular Gaussian noise was introduced to enhance trajectory diversity and authenticity. Additionally, a multi-perspective trajectory shape recognition algorithm was designed, which could effectively integrate image and sequential multi-modal features by adopting a multilayer perceptron, enabling precise trajectory shape recognition. Experimental results demonstrate that the proposed …

Research On Performance Evaluation Of P300 Brain-Computer Interface Under Environment Modeling And Simulation, Xiaofei Ge, Jinling Lian, Jin Han, Xin'an Fan, Danmei Luo, Yuxiang Hua, Hao Liu, Lijian Zhang

Journal of System Simulation

Abstract: In the process of brain-computer interface(BCI) technology stepping from laboratory to practical application scenarios, it is difficult to make an accurate prediction and evaluation of the effect of real environment factors on the system performance. Therefore, a research method based on a multifactor simulation experiment was proposed. A controllable simulation experiment environment was constructed, and the parametric modeling of two key physical environmental factors, namely noise and light, was carried out. By taking the number of electroencephalogram channels as the system parameter, this paper systematically studied the influence mechanism of the aforementioned factors on the decoding performance of P300-BCI. …

Design And Application Of Bds Visual Simulation Teaching Platform Based On Cesium, Liguo Lu, Yulin Cai, Tangting Wu, Canhui Lin

Journal of System Simulation

Abstract: The deep development of the BeiDou navigation satellite system(BDS) cannot be separated from the support of corresponding talent education, especially the development of new digital resources and tools. A visual simulation teaching platform for BDS was designed based on open source Cesium engine library to address issues such as the large and complex size of the BDS project, the unreachable space environment, and the difficult understanding of principles and concepts. The platform adopted a B/S architecture to achieve visual simulation of course content such as orbit cognition, orbit calculation, constellation simulation, service performance, and satellite links. Teaching applications have …

Key Problems Of Intent Recognition Research: A Survey On Activity, Plan And Goal Recognition, Yi Zhang, Kai Xu, Shuilin Li, Dejun Chen, Yunxiu Zeng, Yong Peng

Journal of System Simulation

Abstract: With the development of artificial intelligence technology, realizing intent recognition in human-computer interaction has become one of the key challenges. In this paper, the current research status of three fields was systematically sorted out, namely activity recognition, plan recognition, and goal recognition, and the progress from the problem proposal to the current development was analyzed. The main research approaches in each field were reviewed, and a survey of research on activity recognition, a development overview of plan recognition, and a retrospective analysis of hotspots in goal recognition were conducted. This general view of the problem helped to clarify …

Integrating Geometric Priors And Importance Sampling For High-Fidelity Indoor Scene Reconstruction, Tao Yang, Min Shi, Xigang Zhao, Suqin Wang, Qi Wang, Dengming Zhu

Journal of System Simulation

Abstract: Gaussian splatting suffers from geometric distortion during scene reconstruction, particularly in weakly textured indoor scenes. To address this issue, this paper proposes a high-precision indoor scene reconstruction method that integrates geometric priors and importance sampling. The proposed method fully considers the effect of the initialization process on reconstruction quality. An advanced feed-forward model is employed to generate high-quality geometric initialization, thus improving overall reconstruction stability and accuracy. An importance sampling strategy is introduced to mitigate the adverse effects of blurry images. Furthermore, a supervision mechanism based on a geometric prior model is designed to constrain the scene structure, further …

Research On Gaussian Splatting Modeling Of Power Equipment In 3d Scenes, Haiying Li, Haonan Xu, Junfang Hao

Journal of System Simulation

Abstract: To address the issues of missing camera poses in captured images and poor reconstruction quality in 3D modeling of power equipment, a 3D Gaussian splatting 3D modeling method for power equipment based on video sequences was proposed. Theffmpeg was adopted to extract video frames at a reduced rate, and the Scharr operator was employed to quantify the sharpness of video frames to screen high-quality images for forming an input dataset, ensuring the completeness of equipment poses and the quality of modeling data. Through multi-view feature point extraction and matching, combined with an incremental structure-from-motion algorithm to optimize and …

A Precise Damage Assessment Method For Lethal Blast Warheads Against Quadruped Robots, Xueqian Wang, Jianbing Men, Xin Zhou, Shuyou Wang, Mei Li

Journal of System Simulation

Abstract: To accurately evaluate the damage efficiency of a lethal blast warhead on quadruped robots, a typical quadruped robot replication model and vulnerability damage tree were constructed through Autodesk Inventor. The power field calculation model of a lethal blast warhead was introduced. Based on the high-precision collision detection and graphic rendering technology of UE, this paper carried out the intersection detection of destructive elements and targets and realistic scene visualization. A visualization system for damage assessment of quadruped robots by a lethal blast warhead was developed, featuring capabilities such as parametric modeling of the lethal blast warhead, power field evolution …

Robot Path Planning By Reinforcement Learning Based On Sac3q-Hdm, Dequan Li, Wan Xiong

Journal of System Simulation

Abstract: To address the issues of overestimated and underestimated biases, low sample utilization rate, and the inability to balance exploration and exploitation in reinforcement learning for path planning, an improved SAC method was proposed. The size balance of entropy was explored and utilized through adaptive temperature coefficient adjustment; on the basis of the SAC framework, a triple Critic architecture was introduced to dynamically weight and fuse the minimum and average values through Qvalue uncertainty, balancing overestimated and underestimated biases. A mixed dynamic sampling experience replay buffer was designed; experience data was partitioned based on reward thresholds; sampling ratios were dynamically …