Malware classification of executable files by convolutional networks

US11256804B2 · kind B2 · utility

Assignees

• Avast Software s.r.o. · CZ
• Ustav informatiky AV CR, v.v.i. · CZ

Inventors

• Marek Kr{hacek over (c)}ál · Praha, CZ
• Martin Bálek · Teplice, CZ
• Ond{hacek over (r)}ej {hacek over (S)}vec · Velehrad, CZ
• Martin Vejmelka · Praha, CZ

Key dates

• Filing date: Nov 8, 2018
• Grant date: Feb 22, 2022
• Expiry date: Apr 19, 2040

Classification

• Technology area (CPC G): Physics
• CPC primary: G06F2221/033
• WIPO field: Computer technology
• WIPO sector: Electrical engineering

Abstract

A convolutional deep neural network architecture can detect malicious executable files by reading the raw sequence of bytes, that is, without any domain-specific feature extraction or preprocessing.