Data Security on Mobile Devices: Current State of the Art, Open Problems, and Proposed Solutions

Zinkus, Maximilian; Jois, Tushar M.; Green, Matthew

doi:10.48550/arxiv.2105.12613

Search citation statements

Order By: Relevance

Paper Sections

Select...

Methods1

Citation Types

Supporting

Mentioning

Contrasting

Year Published

2023

Publication Types

Select...

Other1

Relationship

Self Cite0

Independent1

Authors

Journals

Cited by 1 publication

(1 citation statement)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This can be useful for any application where a probability of breakdown by such an attack exists. 1,2 Thus, the solution to the problem comes down to a binary classification, where 1 is a bot and 0 is a human.…”

Section: Methodsmentioning

confidence: 99%

Comparison of the applicability of synergistic models with dense neural networks on the example of mobile device security

Vishnevskaya

Salyp

2023

International Conference on Digital Transformation: Informatics, Economics, and Education (DTIEE2023)

View full text Add to dashboard Cite

The dominant approach in contemporary science and industry applications is deep learning. Deep learning consists of many different architectures, including dense, convolutional and recurrent neural networks. In this article we compare deep learning, particularly dense neural networks (DNNs) with the other approach, synergetic models. As a task to compare the mobile application security was chosen. Two datasets were created, sensor values for human beings interacting with smartphones and malicious bot records using emulators. Deep learning and synergetic models were tasked to distinguish them from each other using binary classification. Both models tackled a problem comparatively well, archiving 99% and 85% accuracy respectively. While the deep learning model performed better, the synergetic model excelled in training speed, versatility and results transparency. Pros and cons of both models were addressed in the results section.

show abstract

Section: Methodsmentioning

confidence: 99%