UCL Discovery
UCL home » Library Services » Electronic resources » UCL Discovery

Machine learning techniques for identification using mobile and social media data

Perez Mila de la Roca, Beatrice M; (2020) Machine learning techniques for identification using mobile and social media data. Doctoral thesis (Ph.D), UCL (University College London). Green open access

[thumbnail of thesis_corrections_20200908.pdf]
Preview
Text
thesis_corrections_20200908.pdf - Accepted Version

Download (3MB) | Preview

Abstract

Networked access and mobile devices provide near constant data generation and collection. Users, environments, applications, each generate different types of data; from the voluntarily provided data posted in social networks to data collected by sensors on mobile devices, it is becoming trivial to access big data caches. Processing sufficiently large amounts of data results in inferences that can be characterized as privacy invasive. In order to address privacy risks we must understand the limits of the data exploring relationships between variables and how the user is reflected in them. In this dissertation we look at data collected from social networks and sensors to identify some aspect of the user or their surroundings. In particular, we find that from social media metadata we identify individual user accounts and from the magnetic field readings we identify both the (unique) cellphone device owned by the user and their course-grained location. In each project we collect real-world datasets and apply supervised learning techniques, particularly multi-class classification algorithms to test our hypotheses. We use both leave-one-out cross validation as well as k-fold cross validation to reduce any bias in the results. Throughout the dissertation we find that unprotected data reveals sensitive information about users. Each chapter also contains a discussion about possible obfuscation techniques or countermeasures and their effectiveness with regards to the conclusions we present. Overall our results show that deriving information about users is attainable and, with each of these results, users would have limited if any indication that any type of analysis was taking place.

Type: Thesis (Doctoral)
Qualification: Ph.D
Title: Machine learning techniques for identification using mobile and social media data
Event: University College London
Open access status: An open access version is available from UCL Discovery
Language: English
Additional information: Copyright © The Author 2020. Original content in this thesis is licensed under the terms of the Creative Commons Attribution 4.0 International (CC BY 4.0) Licence (https://creativecommons.org/licenses/by/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request.
Keywords: Supervised Learning, Classification, Machine Learning, Sensor Data, Identification, Privacy
UCL classification: UCL
UCL > Provost and Vice Provost Offices > UCL SLASH
UCL > Provost and Vice Provost Offices > UCL SLASH > Faculty of S&HS
UCL > Provost and Vice Provost Offices > UCL SLASH > Faculty of S&HS > Dept of Geography
URI: https://discovery.ucl.ac.uk/id/eprint/10109672
Downloads since deposit
100Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item