Explore the privacy implications of combining publicly available social media images with face recognition technology in this Black Hat USA 2011 conference talk. Delve into a series of experiments demonstrating the feasibility of large-scale, automated individual re-identification using profile pictures from popular online social networks. Examine the technical aspects, legal ramifications, and future privacy concerns in an augmented reality world. Learn about facial recognition taxonomies, experimental approaches, and results that highlight the potential for identifying and inferring sensitive information about strangers both online and offline. Discuss key themes, including data accretion, scalability issues, and the challenges of maintaining privacy in the age of advanced facial recognition technologies. Consider the broader implications of these findings and the lack of clear solutions to address the emerging privacy concerns.
Faces of Facebook - Or, How the Largest Real ID Database in the World Came to Be
Overview
Syllabus
Intro
A tale of two futures
Back to the past
Background
of various technologies (2/2)
What is different: The convergence
Our research focus
A short facial taxonomy
In a nutshell
Why this matters
What this implies
Key themes
Experiments
Source DB
Experimenti
Experiment 1: Approach
Experiment 1: Ground truth
Experiment 1: Evaluation
Experiment 1: Results
Experiment 1: Comments
Experiment 2: Process
Experiment 2: Approach
Experiment 2: Examples
Experiment 2: Results
Pushing the envelope
Chances of correctly matching SSN digits by random guess, under status quo knowledge
SSN assignment patterns: Two representative States
Chances of correctly matching SSN digits by random guess, under our algorithm
Predicting SSNs from public data
Can you do 1+1? Experiment 3
A picture is worth a thousand words
Data "accretion"
Not a trivial problem
Privacy in the Age of Augmented Reality
Screenshots
What is in real time, and what is not
Limitations
Extrapolations
Scalability: Availability of images (2/2)
Scalability: Cooperative subjects
Scalability: Geographical restrictions
Implications (2/4)
No clear solution
Taught by
Black Hat
