Navigating the Vast Ocean of Browser Fingerprints

on YouTube ↗34m

" summary

A practical method combines browser fingerprints with feature engineering to build sparse distributed representations that encode similarity directly, using Python/Numenta encoders and composite encoders for version, categorical and periodic fields. Tests use user‑agent parsing (Edgar dataset, a small 115‑row sample) and visualized via multi‑dimensional scaling; SDRs produce tighter clusters and better track device evolution than string‑distance or word2vec approaches. Key data points: HTML5 canvas yields ~18 bits, SDRs avoid constant retraining and are released as open‑source Python for Hadoop use.

AI-written summary · about

" chapters (10)

1. 0:00Introduction and disclaimers
2. 2:13What is browser fingerprinting
3. 8:40Traditional fingerprint use cases
4. 12:39Feature engineering for fingerprints
5. 17:37Sparse distributed representations
6. 20:25How sparse encodings work
7. 27:15Encoding version numbers and composites
8. 29:23Visualizing fingerprint distances
9. 31:56Coolest InfoSec creatures
10. 33:19Q&A and closing remarks

" related talks

• 14mBrowser Fingerprinting: Stalking With a Personal TouchEivind Arvesen · NDC Oslo 2020
• 21mFp-Scanner: The Privacy Implications of Browser Fingerprint InconsistenciesAntoine Vastel, Pierre Laperdrix, Walter Rudametkin, Romain Rouvoy · USENIX Security '18
• 24mTales From the Trenches: Fingerprints on the WebIgor Trindade Oliveira · JSConf EU 2018
• 43mAround the Web in 80 Hours: Scalable Fingerprinting with Chromium AutomationIsaac Dawson · CODE BLUE 2016
• 40mAn Analysis of the Fingerprinting-Scripts and their Actors on the InternetSebastian Neef · code.talks 2022
• 26mk-fingerprinting: A Robust Scalable Website Fingerprinting TechniqueGeorge Danezis, Jamie Hayes · USENIX Security '16
• 35mBreaking fraud & bot detection solutionsMayank Dhiman · OWASP AppSec USA 2018
• 44mApplied Machine Learning for Data Exfil and Other Fun TopicsMatt Wolff, Brian Wallace, Xuan Zhao · Black Hat USA 2016