Analysis of the vulnerability of machine learning regression models to adversarial attacks using data from 5G wireless networks
Leonid Legashev, Artur Zhigalov, Denis Parfenov·May 01, 2025
Summary
Machine learning regression models face vulnerability to adversarial attacks, particularly in 5G networks. An FGSM attack can increase Mean Squared Error (MSE) by 33% and decrease R2 by 10%. LightGBM, optimized for hyperparameters, identifies adversarial anomalies with 98% accuracy, improving R2 by 10% and reducing MSE by 1.95%. This highlights the susceptibility of regression models to attacks, with network analysis crucial for detecting malicious activity. The text also discusses strategies to mitigate the impact of adversarial examples on robust nonparametric models, emphasizing security in automotive self-organizing networks.
Introduction
Background
Overview of machine learning regression models
Importance of 5G networks in modern technology
Objective
To explore the vulnerability of machine learning regression models to adversarial attacks in 5G networks
To analyze the impact of FGSM attacks on model performance metrics
Adversarial Attacks on Machine Learning Regression Models
FGSM Attack Analysis
Description of FGSM attack
Effects on Mean Squared Error (MSE) and R2 score
Vulnerability Assessment
Identification of regression models' susceptibility to adversarial attacks
LightGBM's Role in Adversarial Anomaly Detection
LightGBM Model
Overview of LightGBM
Optimization for hyperparameters
Adversarial Anomaly Detection
Accuracy of LightGBM in identifying adversarial anomalies
Improvement in R2 score and reduction in MSE
Network Analysis for Malicious Activity Detection
Importance of Network Analysis
Role in identifying and mitigating adversarial attacks
Techniques for Malicious Activity Detection
Overview of network analysis methods
Strategies for Mitigating Adversarial Examples
Robust Nonparametric Models
Importance of robustness in nonparametric models
Security in Automotive Self-Organizing Networks
Application of security measures in automotive networks
Conclusion
Summary of Findings
Future Research Directions
Importance of Continuous Security Enhancements
Basic info
papers
cryptography and security
artificial intelligence
Advanced features
Insights
What are the key strategies discussed for mitigating adversarial attacks on regression models in 5G networks?
How does the implementation of security measures in automotive self-organizing networks enhance robustness against adversarial examples?
How does LightGBM optimize hyperparameters to identify adversarial anomalies in 5G networks?
What impact does an FGSM attack have on the performance metrics of regression models in 5G networks?