Pitfalls in Machine Learning for Computer Security.

This article presents ten common pitfalls of machine learning in the context of computer security. Pitfalls are included from all stages of the machine learning process including data snooping, inappropriate baseline, and base rate fallacy. Next, the prevalence of each pitfall was assessed using thi...

Descripción completa

Detalles Bibliográficos
Publicado en:	Communications of the ACM Vol. 67; no. 11; pp. 104 - 113
Autores principales:	Arp, Daniel, Quiring, Erwin, Pendlebury, Feargus, Warnecke, Alexander, Pierazzi, Fabio, Wressnegger, Christian, Cavallaro, Lorenzo, Rieck, Konrad
Formato:	Artículo
Publicado:	Association for Computing Machinery Nov2024
Materias:	Machine learning Computer security Malware Systems design Acquisition of data Intrusion detection systems (Computer security)
Acceso en línea:	Ver este registro en EBSCOhost

Descripción
Sumario:	This article presents ten common pitfalls of machine learning in the context of computer security. Pitfalls are included from all stages of the machine learning process including data snooping, inappropriate baseline, and base rate fallacy. Next, the prevalence of each pitfall was assessed using thirty security papers published in the last ten years. Then, an impact analysis is presented of these pitfalls in four different security fields, including vulnerability and network intrusion detection.