[3075] in Privacy_Forum
[ PRIVACY Forum ] Poison Attacks Against Machine Learning
daemon@ATHENA.MIT.EDU (PRIVACY Forum mailing list)
Sun Jul 22 12:57:10 2012
Date: Sun, 22 Jul 2012 09:41:35 -0700
To: privacy-list@vortex.com
Message-ID: <20120722164135.GB6614@vortex.com>
MIME-Version: 1.0
Content-Disposition: inline
From: PRIVACY Forum mailing list <privacy@vortex.com>
Reply-To: PRIVACY Forum mailing list <privacy@vortex.com>
Content-Type: text/plain; charset="iso-8859-1"
Errors-To: privacy-bounces+privacy-forum=mit.edu@vortex.com
Content-Transfer-Encoding: 8bit
Poison Attacks Against Machine Learning
http://j.mp/LGQrRy (i-programmer)
"Three researchers, Battista Biggio (Italy) Blaine Nelson and Pavel
Laskov (Germany), have found a way to feed an SVM with data specially
designed to increase the error rate of the machine as much as possible
with a few data points. The approach assumes that the attacker knows
the learning algorithm being employed and has access to the same data.
Less realistically it assumes that the attacker has access to the
original training data. This is unlikely, but the original training
data could be approximated by a sample from the population. With all
of the data the attacker can manipulate the optimal SVM solution by
inserting crafted attack points."
- - -
--Lauren--
Lauren Weinstein (lauren@vortex.com): http://www.vortex.com/lauren
Co-Founder: People For Internet Responsibility: http://www.pfir.org
Founder:
- Data Wisdom Explorers League: http://www.dwel.org
- Network Neutrality Squad: http://www.nnsquad.org
- Global Coalition for Transparent Internet Performance: http://www.gctip.org
- PRIVACY Forum: http://www.vortex.com
Member: ACM Committee on Computers and Public Policy
Lauren's Blog: http://lauren.vortex.com
Google+: http://vortex.com/g+lauren / Twitter: http://vortex.com/t-lauren
Tel: +1 (818) 225-2800 / Skype: vortex.com
_______________________________________________
privacy mailing list
http://lists.vortex.com/mailman/listinfo/privacy
