DeepMemIntrospect: Recognizing Data Structures in Memory with Neural Networks

Chung Kuan Chen, E. Lin Ho, Shiuh-Pyng Shieh

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

One yet-To-be-solved but very vital memory forensic problem is to recover data structure information from a specified memory range. Unlike previous studies relying on fixed signature of value or structure, DeepMemlntrospect is the first convolution neural network (CNN) based memory forensic system that can recover data structure information merely from raw memory without relying on signatures. Our experimental results demonstrate high accuracy with over 99% and also show significant performance improvement.

Original languageEnglish
Title of host publicationDSC 2018 - 2018 IEEE Conference on Dependable and Secure Computing
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781538657904
DOIs
StatePublished - 23 Jan 2019
Event2018 IEEE Conference on Dependable and Secure Computing, DSC 2018 - Kaohsiung, Taiwan
Duration: 10 Dec 201813 Dec 2018

Publication series

NameDSC 2018 - 2018 IEEE Conference on Dependable and Secure Computing

Conference

Conference2018 IEEE Conference on Dependable and Secure Computing, DSC 2018
CountryTaiwan
CityKaohsiung
Period10/12/1813/12/18

Keywords

  • Memory forensic
  • data structure reversing
  • deep learning

Fingerprint Dive into the research topics of 'DeepMemIntrospect: Recognizing Data Structures in Memory with Neural Networks'. Together they form a unique fingerprint.

Cite this