Národní úložiště šedé literatury Nalezeno 1 záznamů.  Hledání trvalo 0.00 vteřin. 
Adversarial Attacks on AI Algorithms and Their Prevention
Gregorová, Jana ; Vaško, Marek (oponent) ; Herout, Adam (vedoucí práce)
The trustworthiness of AI, adversarial attacks on AI, and explainability of deep machine learning models represent complex and insufficiently explored topics. This thesis provides a comprehensive overview of state-of-the-art key methods for adversarial attacks on AI in computer vision, their explanation and prevention. By making this topic more accessible and understandable, the work aims to engage a broader audience in research of the security of AI and explainability of AI. Furthermore, this thesis delves into methods for explaining individual classification decisions of deep learning classifiers through Explainable AI (XAI) techniques. It also introduces a tool that integrates different methods for conducting adversarial examples with the application of XAI methods, allowing for monitoring AI attacks and analyzing the decision-making process of deep classifiers during such attacks.

Chcete být upozorněni, pokud se objeví nové záznamy odpovídající tomuto dotazu?
Přihlásit se k odběru RSS.