[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: 27C3 on Tor



Thus spake Roc Admin (onionroutor@xxxxxxxxx):

> Found the presentation at least and watching it now.  You have to skip
> to 3:30 for the actual presentation.
> 
> http://www.youtube.com/watch?v=crMInOosdBk&feature=related

Here's a direct link to the relevant section:
http://www.youtube.com/watch?v=crMInOosdBk#t=31m38s

The work seems to be pretty thorough, but it would be nice to have a
paper and/or the source to their classifier. The 27c3 website doesn't
seem to list much:
http://events.ccc.de/congress/2010/Fahrplan/events/4140.en.html

The results aren't very specific in the video. It looks like they
covered the major pitfall of timing attack research (the Base Rate
Fallacy) with their presentation of the True Positive rate and the
False Positive rate, but it's not 100% clear, because there's only a
single slide with the results:
http://www.youtube.com/watch?v=crMInOosdBk#t=46m20s

The other detail that's not clear is if you train your detector for 5
interesting websites, and 2000 uninteresting ones, what happens if the
target is browsing sites outside of those 2000 uninteresting ones?

-- 
Mike Perry
Mad Computer Scientist
fscked.org evil labs

Attachment: pgpX8AKACipz3.pgp
Description: PGP signature