Improved Feature Engineering for Free-Text Keystroke Dynamics

Eden Abadi, Itay Hazan

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

Abstract

Free-text keystroke dynamics is a method of verifying users’ identity based on their unique pattern of typing a spontaneous text on a keyboard. When applied in remote systems, it can add an additional layer of security that can detect compromised accounts. Therefore, service providers can be more certain that remote systems accounts would not be compromised by malicious attackers. Free-text keystroke dynamics usually involve the extraction of n-graphs, which represent the latency between n consecutive events. These n-graphs are then integrated with one of the various existing machine learning algorithms. To the best of our knowledge, n-graphs are the most widely used feature engineering for free text keystroke dynamics. We present extended-n-graphs, an improved version of the commonly used n-graphs, based on several extended metrics that outperform the traditionally used basic n-graphs. Our technique was evaluated on top of the gradient boosting algorithm, best performing algorithm on basic n-graphs and several additional algorithms such as random forest, K-NN, SVM and MLP. Our empirical results show encouraging 4% improvement in the Area Under the Curve (AUC) when evaluated on a publicly used benchmark.

Original languageEnglish
Title of host publicationSecurity and Trust Management - 16th International Workshop, STM 2020, Proceedings
EditorsKostantinos Markantonakis, Marinella Petrocchi
PublisherSpringer Science and Business Media Deutschland GmbH
Pages93-105
Number of pages13
ISBN (Print)9783030598167
DOIs
StatePublished - 1 Jan 2020
Externally publishedYes
Event16th International Workshop on Security and Trust Management, STM 2020, held in conjunction with the 25th European Symposium on Research in Computer Security, ESORICS 2020 - Guildford, United Kingdom
Duration: 17 Sep 202018 Sep 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12386 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference16th International Workshop on Security and Trust Management, STM 2020, held in conjunction with the 25th European Symposium on Research in Computer Security, ESORICS 2020
Country/TerritoryUnited Kingdom
CityGuildford
Period17/09/2018/09/20

Keywords

  • Feature engineering
  • Free text
  • Keystroke dynamics

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science (all)

Fingerprint

Dive into the research topics of 'Improved Feature Engineering for Free-Text Keystroke Dynamics'. Together they form a unique fingerprint.

Cite this