Continuous Authentication Using Mouse Clickstream Data Analysis

Almalki, Sultan; Chatterjee, Prosenjit

doi:10.1007/978-3-030-24900-7_6

Cited by 25 publications

(7 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Both the 1D-CNN and LSTM-RNN observed miniscule standard deviations among the top-10 users, indicating consistent authentications when model is performing at a high level, but this cannot be extrapolated to all users. Random forests appear to be one of the generally best performing machine learning algorithms for user authentication on mouse dynamics in accordance with the previous literature [34,35], and our machine learning models' results concur. The SVM and KNN had lower performances on the dataset as compared to the random forest.…”

Section: Discussion and Analysissupporting

confidence: 91%

Machine and Deep Learning Applications to Mouse Dynamics for Continuous User Authentication

Siddiqui

Dave

Vanamala

et al. 2022

MAKE

View full text Add to dashboard Cite

Static authentication methods, like passwords, grow increasingly weak with advancements in technology and attack strategies. Continuous authentication has been proposed as a solution, in which users who have gained access to an account are still monitored in order to continuously verify that the user is not an imposter who had access to the user credentials. Mouse dynamics is the behavior of a user’s mouse movements and is a biometric that has shown great promise for continuous authentication schemes. This article builds upon our previous published work by evaluating our dataset of 40 users using three machine learning and three deep learning algorithms. Two evaluation scenarios are considered: binary classifiers are used for user authentication, with the top performer being a 1-dimensional convolutional neural network (1D-CNN) with a peak average test accuracy of 85.73% across the top-10 users. Multi-class classification is also examined using an artificial neural network (ANN) which reaches an astounding peak accuracy of 92.48%, the highest accuracy we have seen for any classifier on this dataset.

show abstract

Section: Discussion and Analysissupporting

confidence: 91%

Machine and Deep Learning Applications to Mouse Dynamics for Continuous User Authentication

Siddiqui

Dave

Vanamala

et al. 2022

MAKE

View full text Add to dashboard Cite

show abstract

“…X is the characteristic matrix of the mouse, which is divided into training set and test set according to different O values. L is the mouse label matrix, and the size of the matrix depends on the number of samples n. is system is a binary classification model [54], so the label is defined as (0, 1), where the legal user is 1 and the illegal user is 0. Op for libsvm training custom parameters, including kernel function and other values, in this system is mainly selected by using the exhaustive method.…”

Section: Training and Testing Of The Mouse Modelmentioning

confidence: 99%

Design and Implementation of Continuous Authentication Mechanism Based on Multimodal Fusion Mechanism

Guan

Zhang

2021

Security and Communication Networks

View full text Add to dashboard Cite

Most of the current authentication mechanisms adopt the “one-time authentication,” which authenticate users for initial access. Once users have been authenticated, they can access network services without further verifications. In this case, after an illegal user completes authentication through identity forgery or a malicious user completes authentication by hijacking a legitimate user, his or her behaviour will become uncontrollable and may result in unknown risks to the network. These kinds of insider attacks have been increasingly threatening lots of organizations, and have boosted the emergence of zero trust architecture. In this paper, we propose a Multimodal Fusion-based Continuous Authentication (MFCA) scheme, which collects multidimensional behaviour characteristics during the online process, verifies their identities continuously, and locks out the users once abnormal behaviours are detected to protect data privacy and prevent the risk of potential attack. More specifically, MFCA integrates the behaviours of keystroke, mouse movement, and application usage and presents a multimodal fusion mechanism and trust model to effectively figure out user behaviours. To evaluate the performance of the MFCA, we designed and implemented the MFCA system and the experimental results show that the MFCA can detect illegal users in quick time with high accuracy.

show abstract

“…They used a one class SVM classifier that achieved a FAR of 0.37% and a FRR of 1.12%. In [2] Almalki et al conducted an empirical evaluation of classical machine learning classification techniques for mouse dynamics in continuous authentication scenarios. They extracted a set of 39 features (inspired in [3]) from individual mouse actions.…”

Section: Related Workmentioning

confidence: 99%

“…Finally, we summarise in Table 5 a comparison of our approach and evaluation results with those of existing works in literature in terms of (1) adversarial strategy used to generate attacks, (2) whether the approach is evaluated in a public dataset,(3) the AUC performance of authentication models, (4) the attack success rate for artificial generated attacks; and (5) whether the generated can be adapted to goal-oriented attacks. The question mark in Goaloriented attack column means that we think those kinds of attacks Session 1: Adversarial Machine Learning AISec '20, November 13, 2020, Virtual Event, USA are extensible to goal-oriented scenarios because of the way they were designed, but there is no formal studies proving their goaloriented feasibility.…”

Section: Related Workmentioning

confidence: 99%

Scrap

Solano¹,

López²,

Rivera³

et al. 2020

Proceedings of the 13th ACM Workshop on Artificial Intelligence and Security

View full text Add to dashboard Cite

Adversarial attacks have gained popularity recently due to their simplicity and impact. Their applicability to diverse security scenarios is however less understood. In particular, in some scenarios, attackers may come up naturally with ad-hoc black-box attack techniques inspired directly on characteristics of the problem space rather than using generic adversarial techniques. In this paper we explore an intuitive attack technique for Mouse-based Behavioral Biometrics and compare its effectiveness against adversarial machine learning attacks. We show that attacks leveraging on domain knowledge have higher transferability when applied to various machine-learning techniques and are also more difficult to defend against. We also propose countermeasures against such attacks and discuss their effectiveness.

show abstract

Continuous Authentication Using Mouse Clickstream Data Analysis

Cited by 25 publications

References 13 publications

Machine and Deep Learning Applications to Mouse Dynamics for Continuous User Authentication

Machine and Deep Learning Applications to Mouse Dynamics for Continuous User Authentication

Design and Implementation of Continuous Authentication Mechanism Based on Multimodal Fusion Mechanism

Scrap

Contact Info

Product

Resources

About