Guardians of the Web: Harnessing Machine Learning to Combat Phishing Attacks

Gamification and Augmented Reality., Journal Year: 2025, Volume and Issue: 3, P. 91 - 91

Published: Jan. 16, 2025

Phishing remains one of the most dangerous threats to internet users and organizations today since it utilizes spoofed websites coax into revealing their data. This paper focuses on effectiveness algorithms in detecting such abusive websites. It goes analyze dataset phishing non- URLs providing explanatory attributes as domain registration date, URL length or existence HTTPS. The models studied include Decision Tree, Random Forest, Support Vector Machines. results found that Forest algorithm had best performance 97% terms classification accuracy, Machines performed generalization accuracy with precision recall values 0.92 0.95, respectively. study investigates feature selection determinants structural features which are crucial determining efficiency detection. Also, enhance model assessment stratified 10-fold cross-validation technique was reduce bias variance. These Results show prospect One Layer Neural Networks a tool improve Detection Systems help provide low-cost fast solutions for current future cyberspace struggles. work aims increase confidence online security applications against modern methods.The proposed modifications will strengthen counter measures attacks shifting technological context while also working towards sustaining thus require further inquiry facets applicability sophisticated artificial intelligence techniques use useful yet diverse sets data incorporation explainable intelligent systems

Language: Английский

---

Phishing Website Detection Using Machine Learning

Gamification and Augmented Reality., Journal Year: 2025, Volume and Issue: 3, P. 81 - 81

Published: Jan. 16, 2025

Phishing attacks continue to be a danger in our digital world, with users being manipulated via rogue websites that trick them into disclosing confidential details. This article focuses on the use of machine learning techniques process identifying phishing websites. In this case, study was undertaken critical factors such as URL extension, age domain, and presence HTTPS whilst exploring effectiveness Random Forest, Gradient Boosting and, Support Vector Machines algorithms allocating status or non-phishing. study, dataset containing real URLs are employed build model using feature extraction. Following this, various were put test dataset; out all models, Forest performed exceptionally well having achieved an accuracy 97.6%, also found extremely effective possessing strong accuracy. we compared discussed methods detect site. Some features affect detection performance include length, special characters focus even more aspects need further development. The new proposed method improves because applied, recall (true positive) increase, while false positive decrease. results enrich electronic security system, they enable time mode. has demonstrated importance employing cutting-edge deal safeguard against advanced cyber threats, thus laying groundwork for innovation systems future

Language: Английский

---

Diabetes Prediction and Management Using Machine Learning Approaches

Data & Metadata, Journal Year: 2025, Volume and Issue: 4, P. 545 - 545

Published: Feb. 18, 2025

Diabetes has emerged as a significant global health issue, especially with the increasing number of cases in many countries. This trend Underlines need for greater emphasis on early detection and proactive management to avert or mitigate severe complications this disease. Over recent years, machine learning algorithms have shown promising potential predicting diabetes risk are beneficial practitioners. Objective: study highlights prediction capabilities statistical non-statistical methods over classification 768 samples from Pima Indians Database. It consists demographic clinical features age, body mass index (BMI) blood glucose levels that greatly depend vulnerability against Diabetes. The experimentation assesses various types terms accuracy effectiveness regarding prediction. These include Logistic Regression, Decision Tree, Random Forest, K-Nearest Neighbors, Naive Bayes, Support Vector Machine, Gradient Boosting Neural Network Models. results show algorithm gained highest predictive 78.57%, then Forest had second position 76.30% accuracy. findings techniques not just highly effective. Still, they also can potentially act screening tools within data-driven fashion valuable information who is more likely get affected. In addition, help realize timely intervention longer term, which step towards reducing outcomes disease burden attributable healthcare systems.

Language: Английский

---

Improving Oral Cancer Outcomes Through Machine Learning and Dimensionality Reduction

Data & Metadata, Journal Year: 2025, Volume and Issue: 3

Published: Jan. 2, 2025

Oral cancer presents a formidable challenge in oncology, necessitating early diagnosis and accurate prognosis to enhance patient survival rates. Recent advancements machine learning data mining have revolutionized traditional diagnostic methodologies, providing sophisticated automated tools for differentiating between benign malignant oral lesions. This study comprehensive review of cutting-edge including Neural Networks, K-Nearest Neighbors (KNN), Support Vector Machines (SVM), ensemble techniques, specifically applied the cancer. Through rigorous comparative analysis, our findings reveal that Networks surpass other models, achieving an impressive classification accuracy 93.6% predicting Furthermore, we underscore potential benefits integrating feature selection dimensionality reduction techniques model performance. These insights significant promise advanced bolstering detection, optimizing treatment strategies, ultimately improving outcomes realm oncology

Language: Английский

---

Machine learning-based prediction and classification of psychiatric symptoms induced by drug and plants toxicity

Gamification and Augmented Reality., Journal Year: 2025, Volume and Issue: 3, P. 107 - 107

Published: Feb. 12, 2025

Psychiatric disorders induced by drug and plant toxicity represent a complex underexplored area in medical research. Exposure to substances such as pharmaceuticals, illicit drugs, environmental toxins can trigger wide range of neuropsychiatric symptoms. This study proposes the development machine learning (ML) model predict classify these symptoms analyzing open-access, de-identified datasets. Supervised unsupervised techniques, including neural networks algorithms like XGBoost, were applied distinguish drug-induced psychiatric conditions from primary disorders. The models evaluated using metrics accuracy, precision, recall, AUC-ROC. XGBoost demonstrated best performance, achieving an AUC-ROC 94.8%, making it promising tool for clinical decision-support systems. approach improve early detection intervention associated with toxicity, contributing safer more personalized healthcare.

Language: Английский

---

A Machine Learning Model for Diagnosis and Differentiation of Schizophrenia, Bipolar Disorder and Borderline Personality Disorder

LatIA, Journal Year: 2025, Volume and Issue: 3, P. 133 - 133

Published: March 20, 2025

Schizophrenia, bipolar disorder, and borderline personality disorder present overlapping symptoms, complicating accurate diagnosis. Misdiagnosis leads to inappropriate treatment, increased patient distress, higher healthcare burdens. This study develops a machine learning model integrating clinical, neuroimaging, behavioral data improve diagnostic accuracy. The utilizes Convolutional Neural Networks (CNNs) for Gradient Boosting Machines (GBMs) structured clinical data, Recurrent (RNNs) speech analysis. combined demonstrated superior accuracy (94.1%) compared individual models. SHAP analysis identified key features, including specific brain regions, cognitive measures, patterns. External validation confirmed robustness, highlighting the model’s potential as decision-support tool. Future research should focus on enhancing interpretability real-time support.

Language: Английский

---

Automated Quantification of Vesicoureteral Reflux using Machine Learning with Advancing Diagnostic Precision

Data & Metadata, Journal Year: 2024, Volume and Issue: 4, P. 460 - 460

Published: Nov. 7, 2024

This article uses machine learning to quantify vesicoureteral reflux (VUR). VCUGs in pediatric urology are used diagnose VUR. The goal is increase diagnostic precision. Various models categorize VUR grades (Grade 1 Grade 5) and evaluated using performance metrics confusion matrices. Study datasets come from internet repositories with repository names accession numbers. Machine performed well across several measures. KNN, Random Forest, AdaBoost, CN2 Rule Induction consistently scored 100% AUC, CA, F1-score, precision, recall, MCC, specificity. These classified individually collectively. In contrast, the Constant model poorly all criteria, suggesting its inability reliably. With most excellent average ratings, excelled at grade categorization. Confusion matrices demonstrate that predict grades. large diagonal numbers of show regularly predicted effectively. However, model's constant 5 forecast reduced differentiation. study shows methods automate measurement. findings aid objective grading radiographic evaluation. accurately classifies learning-based techniques may clinical decision-making, patient outcomes.

Language: Английский

---

Comparative Analysis of Advanced Data Mining Methods for Enhancing Medical Diagnosis and Prognosis

Data & Metadata, Journal Year: 2024, Volume and Issue: 3

Published: Oct. 29, 2024

Accurate and early diagnosis, coupled with precise prognosis, is critical for improving patient outcomes in various medical conditions. This paper focuses on leveraging advanced data mining techniques to address two key challenges: diagnosis prognosis. Diagnosis involves differentiating between benign malignant conditions, while prognosis aims predict the likelihood of recurrence after treatment. Despite significant advances imaging clinical collection, achieving high accuracy both remains a challenge. study provides comprehensive review state-of-the-art machine learning used including Neural Networks, K-Nearest Neighbors (KNN), Naïve Bayes, Logistic Regression, Decision Trees, Support Vector Machines (SVM). These methods are evaluated their ability process large, complex datasets produce actionable insights practitioners.We conducted thorough comparative analysis based performance metrics such as accuracy, Area Under Curve (AUC), precision, recall, specificity. Our findings reveal that Networks consistently outperform other terms diagnostic predictive capacity, demonstrating robustness handling high-dimensional nonlinear data. research underscores potential algorithms revolutionizing effective thus facilitating more personalized treatment plans improved healthcare outcomes.

Language: Английский

---

From Complexity to Clarity: Improving Microarray Classification with Correlation-Based Feature Selection

LatIA, Journal Year: 2024, Volume and Issue: 3, P. 84 - 84

Published: Nov. 30, 2024

Gene microarray classification is yet a difficult task because of the bigness data and limited number samples available. Thus, need for efficient selection subset genes necessary to cut down on computation costs improve performance. Consistently, this study employs Correlation-based Feature Selection (CFS) algorithm identify informative genes, thereby decreasing dimensions isolating discriminative features. Thereafter, three classifiers, Decision Table, JRip OneR were used assess The strategy was implemented eleven such that reduced compared with complete gene set results. observed results lead conclusion CFS efficiently eliminates irrelevant, redundant, noisy features as well. This method showed great prediction opportunities relevant differentiation datasets. performed best among Table by average accuracy in all mentioned However, approach has many advantages enhances several classes large numbers high time complexity.

Language: Английский

---

Real-Time UAV Recognition Through Advanced Machine Learning for Enhanced Military Surveillance

Gamification and Augmented Reality., Journal Year: 2024, Volume and Issue: 3, P. 63 - 63

Published: Dec. 3, 2024

In an era where the military utilization of Unmanned Aerial Vehicles (UAVs) has become essential for surveillance and operational operations, our study tackles growing demand real-time, accurate UAV recognition. The rise UAVs presents numerous safety hazards, requiring systems that distinguish from non-threatening phenomena, such as birds. This research conducts a comparative examination advanced machine learning models, aiming to address challenge real-time aerial classification in diverse environmental conditions without model retraining. employs extensive datasets train validate models Neural Networks, Support Vector Machines, ensemble methods, Gradient Boosting Machines. fashions are evaluated based on accuracy, forgetfulness, processing efficiency—criteria determining viability scenarios. findings indicate Networks exhibit enhanced performance, demonstrating exceptional accuracy distinguishing culminates primary assertion: possess vital security ramifications can markedly enhance allocation defense resources. significantly improve systems, highlighting effectiveness machine-learning methods identification. Moreover, incorporating Network into defenses is recommended decision-making capabilities operations. Foresee forthcoming developments advocate regular updates keep up with increasingly nimble perhaps stealthier drone designs.

Language: Английский

---