August 7, 2025Security and PrivacyOpen Access

A Forensic Analysis Framework for Machine Learning Model Poisoning Detection

View Full Paper

Ask AI

Bookmark

The framework effectively detects model poisoning in machine learning by analyzing internal model behavior.
It integrates structural, statistical, and interpretability-driven methods to reveal manipulation signals.
Evaluation includes diverse machine learning models like ResNet-18 and YOLOv8, ensuring broad applicability.
Testing on CIFAR-10 and CelebA datasets shows high detection accuracy and valuable forensic insights.

Authors

Discussion

No takes yet. Share an insight, caveat, or question.

Framework detects model poisoning in machine learning systems using forensics, indicating scalable defenses.

Monkam et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark