NobleBlocks

Research indexBook chapter

Defensive Preprocessing Techniques

Trajkovski

Apress eBooks1 Jan 2026

NobleIDark:/48914/w1/20261105/C13AA17FDOI10.1007/979-8-8688-2308-4_7

Paywalled

The full text is at the publisher

You can read the abstract here, or open it at the publisher.

Read at the publisher

Abstract

Production artificial intelligence (AI) systems face adversarial examples that exploit precise mathematical vulnerabilities in model decision boundaries, requiring perturbations crafted with exact pixel or signal values to maintain effectiveness. Traditional defensive approaches focus on improving model robustness through retraining, but preprocessing defenses intercept threats before they reach vulnerable models by transforming inputs in ways that disrupt adversarial perturbations while preserving legitimate content.

Discussion

More

Suggest a correction or report