Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
User Profiling through Zero-Permission Sensors and Machine Learning
With the rise of mobile and pervasive computing, users are often ingesting content on the go. Services are constantly competing for attention in a very crowded field. It is only logical that users would allot their attention to the services that are most likely to adapt to their needs and interests....
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
AUC Knowledge Fountain
2023
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| Summary: | With the rise of mobile and pervasive computing, users are often ingesting content on the go. Services are constantly competing for attention in a very crowded field. It is only logical that users would allot their attention to the services that are most likely to adapt to their needs and interests. This matter becomes trivial when users create accounts and explicitly inform the services of their demographics and interests. Unfortunately, due to privacy and security concerns, and due to the fast nature of computing today, users see the registration process as an unnecessary hurdle to bypass, effectively refusing to provide services with personalization information. In other cases, they may provide inaccurate profile information, either due to lack of accuracy, or for malicious purposes. In this thesis, we use machine learning with zero-permission sensors to test the degree to which it can be used to effectively profile a user without necessitating any explicit input. We do so through first iterating through building an application that collects data from the following zero-permission sensors: the gyroscope, accelerometer, and ambient light sensor. Following that, we pass the data through a multi-step transformation process for feature selection, filtration, and homogenization. We then pass this processed training data through machine learning algorithms, enabling accurate user profiling without the need for explicit information gathering. We additionally test the minimum timespan needed to accurately profile a user, and test three machine learning models. We find that it is indeed possible to accurately predict the biological gender of a user, given 1-day intervals, and using a support vector machine. |
|---|