Seeking a project partner: Privacy preserving data mining platform

[cs294 s11]

I am posting this on behalf of Prashanth.

-------------------

Hello,

I am a graduate student in the LoCal project (which focuses on using
computer science technology in the electricity grid). I am interested
in working on privacy preserving data mining for the course project. I
have still not hooked up with anyone for the project and was wondering
if someone is interested in this idea.

The idea is to build a privacy preserving data mining platform.
Programmers should be able to write "programs" (not just SQL) to query
and process sensitive data in a transparent fashion as if working on a
central database. I personally like the idea of the users having
control over their data. So if we choose to build upon a model where
the users individually manage their data, such a program submitted to
a platform should be able to automatically partition the segments of
the code that accesses sensitive data, partition the operations across
many databases, run the program in a sandbox on the local databases,
perturb the final data and aggregate the results back for the original
program.

The idea is still in a nascent state. I think this project will mainly
draw from literature on distributed compilers, differential privacy
and databases. Do let me know if you are still looking for a project
and would like to work together on this.

Thank you,
--
Prashanth Mohan
http://www.cs.berkeley.edu/~prmohan/