[Busec] busec this week: Raef Bassily (Wed 10AM)

Sharon Goldberg goldbe at cs.bu.edu
Tue Oct 8 23:35:40 EDT 2013


Just a reminder for seminar tomorrow with Raef Bassily, who will tell us
about coupled-worlds privacy. Seminar and lunch will be at the usual time
(Wednesday 10AM) and abstract is below.

The following week Bhavana Kanukurthi from UCLA will tell us about
locally-updatable codes.   The talk will be at an unusual time -- Tuesday
at 10AM (which is a "Monday schedule" at BU).

  Sharon

 BUsec Calendar:  http://www.bu.edu/cs/busec/
 BUsec Mailing list:  http://cs-mailman.bu.edu/mailman/listinfo/busec
 How to get to BU from MIT:  Try the CT2 bus or MIT's "Boston Daytime
 Shuttle" http://web.mit.edu/facilities/transportation/shuttles/
daytime_boston.html


Title: Coupled-Worlds Privacy: Exploiting Adversarial Uncertainty in
Statistical Data Privacy
Speaker:  Raef Bassily, Penn State University.
Wednesday Oct 2, 10AM
MCS137

 Abstract:
In this talk, I will present a new framework for defining privacy in
statistical databases that enables reasoning about and exploiting
adversarial uncertainty about the data. Roughly, our framework requires
indistinguishability of the real world in which a mechanism is computed
over the real dataset, and an ideal world in which a simulator outputs some
function of a “scrubbed” version of the dataset (e.g., one in which an
individual user’s data is removed). In each world, the underlying dataset
is drawn from the same distribution in some class (specified as part of the
definition), which models the adversary’s uncertainty about the dataset.

I will argue that our framework provides meaningful guarantees in a broader
range of settings as compared to previous efforts to model privacy in the
presence of adversarial uncertainty. I will also present several natural,
“noiseless” mechanisms that satisfy our definitional framework under
realistic assumptions on the distribution of the underlying data.

Joint work with Adam Groce, Jonathan Katz, and Adam Smith, appearing in
FOCS 2013
****
Title: Locally Updatable and Locally Decodable Codes
Speaker: Bhavana Kanukurthi, UCLA
Tuesday October 8, 2013, 10AM
MCS137

Abstract:
We introduce the notion of locally updatable and locally decodable codes
(LULDCs). In addition to having low decode locality, such codes allow us to
update a codeword (of a message) to a codeword of a different message, by
rewriting just a few symbols. While, intuitively, updatability and
error-correction seem to be contrasting goals, we show that for a suitable,
yet meaningful, metric (which we call the Prefix Hamming metric), one can
construct such codes. Informally, the Prefix Hamming metric allows the
adversary to arbitrarily corrupt bits of the codeword subject to one
constraint -- he does not corrupt more than a $\delta$ fraction of the $t$
``most-recently changed" bits of the codeword (for all $1 \leq t \leq n$,
where $n$ is the length of the codeword).

Our results are as follows. First, we construct binary LULDCs for messages
in ${0,1}^k$ with constant rate, update locality of $O(log^2 k)$, and read
locality of $O(k^\epsilon)$ for any constant $\epsilon<1$. Next, we
consider the case where the encoder and decoder share a secret state and
the adversary is computationally bounded. Here too, we obtain local
updatability and decodability for the Prefix Hamming metric. Furthermore,
we also ensure that the local decoding algorithm never outputs an incorrect
message -- even when the adversary can corrupt an arbitrary number of bits
of the codeword. We call such codes locally updatable locally
decodable-detectable codes (LULDDCs) and obtain dramatic improvements in
the parameters (over the information-theoretic setting). Our codes have
constant rate, an update locality of $O(log k)$ and a read locality of
$O(\lambda log^2 k)$, where $\lambda$ is the security parameter.

Finally, we show how our techniques apply to the setting of dynamic proofs
of retrievability (DPoR) and present a construction of this primitive with
better parameters than existing constructions. In particular, we construct
a DPoR scheme with linear storage, $O(log k)$ write complexity, and
$O(\lambda log k)$ read and audit complexity.

This is joint work with Nishanth Chandran and Rafail Ostrovsky.
<http://www.cs.bu.edu/~goldbe>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs-mailman.bu.edu/pipermail/busec/attachments/20131008/60f915d8/attachment.html>


More information about the Busec mailing list