Don't Lie to Me: Avoiding Malicious Explanations With STEALTH

Don't Lie to Me: Avoiding Malicious Explanations With STEALTH

STEALTH is a method for using some artificial intelligence-generated models without suffering from malicious attacks or associated unfairness issues. STEALTH asks so few queries (one per data cluster) that malicious algorithms cannot detect its operation or know when to lie.

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE software 2023-05, Vol.40 (3), p.43-53
Hauptverfasser: Alvarez, Lauren, Menzies, Tim
Format: Artikel
Sprache:eng
Schlagworte:
Algorithms
Artificial intelligence
Clustering algorithms
Malware
Software algorithms
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:STEALTH is a method for using some artificial intelligence-generated models without suffering from malicious attacks or associated unfairness issues. STEALTH asks so few queries (one per data cluster) that malicious algorithms cannot detect its operation or know when to lie.
ISSN:0740-7459
1937-4194
DOI:10.1109/MS.2023.3244713