Source Code Analysis With Deep Neural Networks DOI
Rebet Jones

Advances in information security, privacy, and ethics book series, Год журнала: 2024, Номер unknown, С. 355 - 378

Опубликована: Дек. 6, 2024

In recent years, deep learning techniques have garnered considerable attention for their effectiveness in identifying vulnerable code patterns with high precision. Nevertheless, leading models such as Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks require extensive computational resources, resulting overhead that poses challenges real-time deployment. This study presents VulDetect, an innovative transformer-based framework vulnerability detection, developed by fine-tuning a pre-trained large language model (GPT) on variety of benchmark datasets containing code. Our empirical analysis demonstrates VulDetect achieves detection accuracy up to 92.65%, surpassing SyseVR VulDeBERT, two the most advanced existing software vulnerabilities.

Язык: Английский

Understanding the Dynamics of Ocean Wave-Current Interactions Through Multivariate Multi-Step Time Series Forecasting DOI Creative Commons
Zaharaddeen Karami Lawal, Hayati Yassin,

Daphne Teck Ching Lai

и другие.

Applied Artificial Intelligence, Год журнала: 2024, Номер 38(1)

Опубликована: Авг. 19, 2024

Understanding ocean wave-current interactions' complex dynamics is crucial for coastal engineering, marine operations, and climate research applications. This study introduces a pioneering data-driven approach by employing advanced deep learning techniques, specifically Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM) models, to forecast both wave current parameters at varying depths. The models are designed capture the temporal relationships inherent in dynamics, considering speed direction, speed, direction as multivariate time series inputs. Two comprehensive experiments conducted, one utilizing historical values of all another focusing on using parameters. Model performance rigorously evaluated across horizons 5, 12, 24 hours ahead metrics such Mean Absolute Error (MAE), Squared (MSE), Root (RMSE). BiLSTM emerges superior model, demonstrating lower errors, particularly higher depths, while nearshore predictions reveal challenges shallower waters. Furthermore, methodology incorporates hyperparameter optimization cross-validation techniques enhance model's robustness. Ultimately, this work represents transformative leap toward smarter oceans, emphasizing fusion fluid bathymetry advance our understanding coupled dynamics. results showcase high accuracy reliability various signifying method's potential applications oceanography, hydrodynamics, renewable energy.

Язык: Английский

Процитировано

4
Screening for Anomalous Safety Condition Among Existing Buildings Using Explainable Machine Learning DOI Creative Commons
Jie Liu, Guiwen Liu, Neng Wang

и другие.

Structural Control and Health Monitoring, Год журнала: 2025, Номер 2025(1)

Опубликована: Янв. 1, 2025

To ensure a safe environment for occupants, evaluating the physical status and service performance of existing buildings is essential. However, large‐scale building condition assessment usually relies on expertise judgment inspectors, which can be costly laborious due to unclear priorities, ambiguous procedures, ineffective operations. address these challenges, this study proposes an explainable machine learning‐based screening model anomalous safety among buildings, narrowing down scope requiring further detailed inspection monitoring. Initially, imbalanced dataset 18,090 survey reports unsafe labels collected. Then, synthetic minority oversampling technique (SMOTE) conducted balance dataset. Subsequently, seven learning models are trained utilizing 10‐fold cross‐validation with grid search. Findings reveal that, based balanced dataset, ensemble significantly better than that individual models. Specifically, XGBoost achieves highest performance, macro‐F1 98.49%, G‐mean value accuracy 98.49%. The final predictive (the SMOTE‐based model) explained using SHapley Additive exPlanations (SHAP). Service year, structure, location three most important features influencing structural safety. This represents promising approach automated optimizing resource allocation, enhancing effectiveness in decision‐making construction maintenance.

Язык: Английский

Процитировано

0
Automatic Feature Selection for Imbalanced Echocardiogram Data Using Event-Based Self-Similarity DOI Creative Commons
Huang‐Nan Huang, Hongmin Chen, Wei‐Wen Lin

и другие.

Diagnostics, Год журнала: 2025, Номер 15(8), С. 976 - 976

Опубликована: Апрель 11, 2025

Background and Objective: Using echocardiogram data for cardiovascular disease (CVD) can lead to difficulties due imbalanced datasets, leading biased predictions. Machine learning models enhance prognosis accuracy, but their effectiveness is influenced by optimal feature selection robust classification techniques. This study introduces an event-based self-similarity approach automatic data. Critical features correlated with progression were identified leveraging patterns. used dataset, visual presentations of high-frequency sound wave signals, patients heart who are treated using three treatment methods: catheter ablation, ventricular defibrillator, drug control-over the course years. Methods: The dataset was classified into nine categories Recursive Feature Elimination (RFE) applied identify most relevant features, reducing model complexity while maintaining diagnostic accuracy. models, including XGBoost CATBoost, trained evaluated. Results: Both achieved comparable accuracy values, 84.3% 88.4%, respectively, under different normalization To further optimize performance, combined a voting ensemble, improving predictive Four essential features-age, aorta (AO), left (LV), atrium (LA)-were as critical found in Random Forest (RF)-voting ensemble classifier. results underscore importance techniques handling robustness, bias automated systems. Conclusions: Our findings highlight potential machine learning-driven analysis patient care providing accurate, data-driven assessments.

Язык: Английский

Процитировано

0
Automated Evidence Collection and Analysis Using AI DOI
Luay Albtosh

Advances in information security, privacy, and ethics book series, Год журнала: 2024, Номер unknown, С. 143 - 186

Опубликована: Дек. 6, 2024

The integration of Artificial Intelligence (AI) in forensic investigations has transformed evidence collection and analysis, enabling quicker more accurate assessments. This chapter examines the evolution application automated AI-driven tools collecting, processing, analyzing digital evidence. AI-based systems assist experts by autonomously identifying, organizing, categorizing massive datasets, thus accelerating traditional investigative workflows. Key AI methods discussed include machine learning algorithms for data classification, natural language processing document computer vision image video recognition. Additionally, we explore implications these technologies legal system, privacy concerns, accuracy reliability derived through automation. also considers potential challenges, such as integrity bias, well future trends applications within forensics.

Язык: Английский

Процитировано

2
Combatting Deepfakes DOI

Ngozi Tracy Aleke

Advances in information security, privacy, and ethics book series, Год журнала: 2024, Номер unknown, С. 375 - 412

Опубликована: Окт. 18, 2024

As we delve into deepfake technology's threat to digital media integrity, it becomes clear that a holistic approach intertwines ethical guidelines with advanced technological solutions is necessary for an effective defense approach. Credence must be given the crucial balance between innovations detection and establishment of rigorous standards, further analysis underscores importance safeguarding privacy, truthfulness, public trust, while advocating continuous innovation, policy evolution, collective efforts thwart potential independent personal harms posed by deepfakes.

Язык: Английский

Процитировано

1
Digital Forensic Data Mining and Pattern Recognition DOI
Luay Albtosh

Advances in information security, privacy, and ethics book series, Год журнала: 2024, Номер unknown, С. 245 - 294

Опубликована: Дек. 6, 2024

Digital forensic data mining and pattern recognition are essential components in enhancing cybersecurity measures practices. This chapter explores the intersection of artificial intelligence digital forensics, emphasizing methodologies technologies that enable extraction meaningful patterns from vast datasets. By leveraging advanced machine learning algorithms, investigators can identify anomalies, classify behaviors, predict potential threats real-time. The integration AI enhances efficiency accuracy investigations, ultimately leading to improved decision-making threat mitigation strategies. Case studies illustrate practical applications these techniques various domains, underscoring transformative forensics.

Язык: Английский

Процитировано

1
The Role of Cybersecurity Legislation in Promoting Data Privacy DOI

Ngozi Tracy Aleke

Advances in information security, privacy, and ethics book series, Год журнала: 2024, Номер unknown, С. 205 - 244

Опубликована: Дек. 6, 2024

The operation of robust cybersecurity legislation plays a fundamental role in safeguarding data privacy an increasingly unified digital terrain, providing legal framework that sensitizes individuals on their rights and privileges as it relates to the protection data, regulates handlers by stipulating applicable rules regulations when handling information subject, establishes enforceable measures breach occurs, thereby fostering culture trust accountability landscape. Cybersecurity covering requirements is critical ensuring safety security every individual's personal information.

Язык: Английский

Процитировано

0
Advances in Cybersecurity and AI: Integrating Machine Learning, IoT, and Smart Systems for Resilience and Innovation Across Domains DOI Creative Commons

Yara Shamoo

World Journal of Advanced Research and Reviews, Год журнала: 2024, Номер 23(2), С. 2450 - 2461

Опубликована: Авг. 30, 2024

Artificial Intelligence (AI) is revolutionizing cybersecurity by offering enhanced threat detection, predictive analytics, and automated responses. However, AI also introduces significant challenges, including bias, lack of transparency, vulnerability to adversarial attacks. This paper examines the dual role in cybersecurity, providing a comprehensive analysis its benefits drawbacks, discussing future securing digital environments.

Язык: Английский

Процитировано

0
Comparative Analysis of LLMs vs. Traditional Methods in Vulnerability Detection DOI

Yara Shamoo

Advances in information security, privacy, and ethics book series, Год журнала: 2024, Номер unknown, С. 335 - 374

Опубликована: Окт. 18, 2024

In the evolving landscape of cybersecurity, detection software vulnerabilities is paramount for ensuring system integrity and protection. This chapter provides a comparative analysis large language models (LLMs) versus traditional methods in vulnerability detection. It explores strengths limitations each approach, focusing on accuracy, efficiency, adaptability, scalability. By examining real-world case studies experimental results, highlights transformative potential LLMs detecting complex vulnerabilities. also discusses implications integrating into existing security frameworks challenges posed by their adoption. serves as guide practitioners researchers seeking to optimize an increasingly dynamic threat environment.

Язык: Английский

Процитировано

0
Source Code Analysis With Deep Neural Networks DOI
Rebet Jones

Advances in information security, privacy, and ethics book series, Год журнала: 2024, Номер unknown, С. 355 - 378

Опубликована: Дек. 6, 2024

In recent years, deep learning techniques have garnered considerable attention for their effectiveness in identifying vulnerable code patterns with high precision. Nevertheless, leading models such as Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks require extensive computational resources, resulting overhead that poses challenges real-time deployment. This study presents VulDetect, an innovative transformer-based framework vulnerability detection, developed by fine-tuning a pre-trained large language model (GPT) on variety of benchmark datasets containing code. Our empirical analysis demonstrates VulDetect achieves detection accuracy up to 92.65%, surpassing SyseVR VulDeBERT, two the most advanced existing software vulnerabilities.

Язык: Английский

Процитировано

0