Machine Learning Methods for Behavioural Biometrics- ON-372

The objective of this research is to create a data architecture and a state-of-the-art machine learning algorithms to build a robust user-profile system that (i) extracts, stores, builds, and analyses synchronously up to 1 million user profiles generating at least 50 behavioral data (alpha-numeric value of 64 bytes) per second, (ii) provides over 5 millions user-profile recognitions per day through predictive modeling and REST API call, (iii) authenticates continuously to detect suspicious activities and anomalies without using cookies, location, and hardware information, and (iv) tolerates effectively the behavioral data noises caused by the modification of input or device such as a mouse, keyboard, mobile or laptop. With the standard web browsers and I/O bandwidth, F8th IDaaS technology collects up to (50) behaviors data per second per client, and in the coming years, F8th IDaaS will support millions of users simultaneously.

Faculty Supervisor:

Abdelkader Ouda

Student:

Sanket Salunke

Partner:

F8th AI

Discipline:

Engineering - computer / electrical

Sector:

Information and cultural industries

University:

Western University