Multi-Stage Dual-Perturbation Attack Targeting Transductive SVMs and the Corresponding Adversarial Training Defense Mechanism DOI Open Access
Li Liu, Haiyan Chen, Changchun Yin

et al.

Electronics, Journal Year: 2024, Volume and Issue: 13(24), P. 4984 - 4984

Published: Dec. 18, 2024

The Transductive Support Vector Machine (TSVM) is an effective semi-supervised learning algorithm vulnerable to adversarial sample attacks. This paper proposes a new attack method called the Multi-Stage Dual-Perturbation Attack (MSDPA), specifically targeted at TSVMs. MSDPA has two phases: initial samples are generated by arbitrary range attack, and finer attacks performed on critical features induce TSVM generate false predictions. To improve TSVM’s defense against MSDPAs, we incorporate training into loss function minimize of both standard during process. improved considers samples’ effect enhances model’s robustness. Experimental results several datasets show that our proposed defense-enhanced (adv-TSVM) performs better in classification accuracy robustness than native other baseline algorithms, such as S3VM. study provides solution capability kernel methods setting.

Language: Английский

Modeling of Forest Fire Risk Areas of Amazonas Department, Peru: Comparative Evaluation of Three Machine Learning Methods DOI Open Access
Alex J. Vergara, Sivmny V. Valqui-Reina, Dennis Cieza-Tarrillo

et al.

Forests, Journal Year: 2025, Volume and Issue: 16(2), P. 273 - 273

Published: Feb. 5, 2025

Forest fires are the result of poor land management and climate change. Depending on type affected eco-system, they can cause significant biodiversity losses. This study was conducted in Amazonas department Peru. Binary data obtained from MODIS satellite occurrence between 2010 2022 were used to build risk models. To avoid multicollinearity, 12 variables that trigger selected (Pearson ≤ 0.90) grouped into four factors: (i) topographic, (ii) social, (iii) climatic, (iv) biological. The program Rstudio three types machine learning applied: MaxENT, Support Vector Machine (SVM), Random (RF). results show RF model has highest accuracy (AUC = 0.91), followed by MaxENT 0.87) SVM 0.84). In fire map elaborated with model, 38.8% region possesses a very low occurrence, 21.8% represents high-risk level zones. research will allow decision-makers improve forest Amazon prioritize prospective strategies such as installation water reservoirs areas zone. addition, it support awareness-raising actions among inhabitants at greatest so be prepared mitigate control generate solutions event occurring under different scenarios.

Language: Английский

Citations

1
Fire classification and detection in imbalanced remote sensing images using a three-sphere model combined with YOLOv5 DOI

Zidong Nie,

Yitian Xu, Jie Zhao

et al.

Applied Soft Computing, Journal Year: 2025, Volume and Issue: 177, P. 113192 - 113192

Published: May 1, 2025

Language: Английский

Citations

0
A temporal perspective on the reliability of wildfire hazard assessment based on machine learning and remote sensing data DOI
Zakaria Matougui, Mohamed Zouidi

Earth Science Informatics, Journal Year: 2024, Volume and Issue: 18(1)

Published: Dec. 10, 2024

Language: Английский

Citations

2
Prediction of Forest-Fire Occurrence in Eastern China Utilizing Deep Learning and Spatial Analysis DOI Open Access
Jing Li, Duan Huang, Chuxiang Chen

et al.

Forests, Journal Year: 2024, Volume and Issue: 15(9), P. 1672 - 1672

Published: Sept. 23, 2024

Forest fires are a major natural calamity that inflict substantial harm on forest resources and the socio-economic landscape. The eastern region of China is particularly susceptible to frequent fires, characterized by high population density vibrant economic activities. Precise forecasting in this area essential for devising effective prevention strategies. This research utilizes blend kernel analysis, autocorrelation standard deviation ellipse method, augmented geographic information systems (GISs) deep-learning techniques, develop an accurate prediction system forest-fire occurrences. model incorporates data meteorological conditions, topography, vegetation, infrastructure, socio-cultural factors produce monthly forecasts assessments. approach enables identification spatial patterns temporal trends fire occurrences, enhancing both precision breadth predictions. results show global local analyses reveal high-incidence areas mainly concentrated Guangdong, Fujian, Zhejiang provinces, with cities like Jiangmen exhibiting distinct concentration characteristics varied distribution Kernel analysis further pinpoints high-density zones primarily Meizhou, Qingyuan, Guangdong Province, Dongfang City Hainan Province. Standard centroid shift indicate significant northward fire-occurrence over past 20 years, expanding range, decreasing flattening, relatively stable direction. performs effectively validation set, achieving accuracy 80.6%, F1 score 81.6%, AUC 88.2%, demonstrating its practical applicability. Moreover, zoning reveals spring winter Zhejiang, Hainan, while autumn shows widespread medium-incidence areas, summer presents lower occurrences most regions. These findings illustrate influence seasonal climate variations highlight necessity enhanced monitoring measures tailored different seasons.

Language: Английский

Citations

1
Enhanced landslide susceptibility mapping in data-scarce regions via unsupervised few-shot learning DOI
L. S. L. Kong, Wenkai Feng, Xiaoyu Yi

et al.

Gondwana Research, Journal Year: 2024, Volume and Issue: 138, P. 31 - 46

Published: Nov. 4, 2024

Language: Английский

Citations

1
Multi-Stage Dual-Perturbation Attack Targeting Transductive SVMs and the Corresponding Adversarial Training Defense Mechanism DOI Open Access
Li Liu, Haiyan Chen, Changchun Yin

et al.

Electronics, Journal Year: 2024, Volume and Issue: 13(24), P. 4984 - 4984

Published: Dec. 18, 2024

The Transductive Support Vector Machine (TSVM) is an effective semi-supervised learning algorithm vulnerable to adversarial sample attacks. This paper proposes a new attack method called the Multi-Stage Dual-Perturbation Attack (MSDPA), specifically targeted at TSVMs. MSDPA has two phases: initial samples are generated by arbitrary range attack, and finer attacks performed on critical features induce TSVM generate false predictions. To improve TSVM’s defense against MSDPAs, we incorporate training into loss function minimize of both standard during process. improved considers samples’ effect enhances model’s robustness. Experimental results several datasets show that our proposed defense-enhanced (adv-TSVM) performs better in classification accuracy robustness than native other baseline algorithms, such as S3VM. study provides solution capability kernel methods setting.

Language: Английский

Citations

0