Kernel machines show greater native robustness than neural networks against adversarial attacks

A new research paper provides a comprehensive analysis of the adversarial robustness of kernel machines. The study empirically investigates the performance of various kernel-based models, including random feature classifiers and Recursive Feature Machines, against both white-box and black-box evasion attacks. The findings suggest that kernel machines exhibit greater native robustness compared to neural networks, with their performance degrading more gradually when subjected to adversarial perturbations. AI

IMPACT Suggests kernel machines may offer a more robust alternative to neural networks in scenarios vulnerable to adversarial attacks.

RANK_REASON Research paper published on arXiv detailing empirical analysis of model robustness. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• adversarial example
• adversarial training
• binary classification
• Black-box Attacks Against Neural Binary Function Detection
• Image datasets of cocoa beans for taxonomy nuances evaluation
• kernel machine
• machine learning
• multiclass classification
• Neural Networks
• random feature classifiers
• Recursive Feature Machines
• tabular datasets
• white-box attacks

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →