Malware Detection via Graph Based Access Behavioral Description and Semi-supervised Learning
| Autor: | Weixuan Mao, Minle Wang, Subing Liu, Zhihui Zhao |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Computer science
business.industry 05 social sciences Graph based 02 engineering and technology Semi-supervised learning Root cause Machine learning computer.software_genre 0202 electrical engineering electronic engineering information engineering Graph (abstract data type) Malware 020201 artificial intelligence & image processing The Internet Artificial intelligence False positive rate 0509 other social sciences 050904 information & library sciences business True positive rate computer |
| Zdroj: | Advances in Intelligent Systems and Computing ISBN: 9783030002138 |
| DOI: | 10.1007/978-3-030-00214-5_153 |
| Popis: | Malicious code is the root cause of many security incidents and still the major threats for the Internet. Understandings on the access behaviors of programs provide ways of malware detection. In this paper, we propose a graph based representations for access behaviors of programs. With similarity metrics on the access behavior graph, we employ a semi-supervised learning algorithm to infer the intent of the programs. The promising result, 98.8% true positive rate at 0.5% false positive rate, shows the ability of our technique on malware detection and the benefits of the access behavior graph. Ability of our technique on malware detection and the benefits of the access behavior graph. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|