[PET] Privacy vs. Machine Learning (Forum in Dresden Sept 21/22)

Thorsten Strufe strufe.pub at googlemail.com
Tue Apr 21 15:34:46 BST 2015

Hi everybody,

considering the improvements and ubiquity of machine learning
applications, and the impact they (may) have on the privacy of
individuals, we want to bring together people interested in either of
the two sides, here in Dresden on Sept 21/22 2015 ( http://www.prinf.eu ).
To ignite discussions, we see two (rather obvious) topics, that we think
are important and interesting:
1)  considering all the public personal data by all sorts of people, we
wonder how good */inference attacks/* can actually get, even on people
who don't publish (a lot of) information about themselves
  - directly related of course is the question, if we can find new
*/privacy metrics,/* and
- if there are ways to encounter such attacks, without diminishing the
utility for the users (too much). And

2) on a related note we want to further explore ideas towards
*/privacy-preserving recommenders/*.

We will organize this event as a "traditional" scientific workshop
soliciting submissions, which will be reviewed and subsequently
published, to make it easier for the academic audiences to convince
their funding entity of its importance ;-) - but we really mainly want
to engage in discussions, may be fostering some future collaborations,
as well.

I'm attaching the usual Call for Papers - and I hope, of course, that we
will receive some interesting submissions (to make the official part
interesting - so help in advertising is appreciated). We will accept
both novel scientific contributions, but also datasets and replication
studies. But most importantly, we hope to attract a broad audience of
interested participants from different professional backgrounds, to
facilitate great discussions!

Thanks a lot!

Ulf & Thorsten

Thorsten Strufe                   
TU Dresden                           https://dud.inf.tu-dresden.de/
CASED                                          http://www.cased.de/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.links.org/pipermail/pet/attachments/20150421/94c3e7ca/attachment.html>
-------------- next part --------------
Call for Papers:

Workshop on Privacy and Inference (PrInf 2015)

TU Dresden, September, 21/22 (TBA)


The First Workshop on Privacy and Inference will be held in conjunction
with the 38th German Conference on Artificial Intelligence that will
take place in September 21-25, 2015 in Dresden, Germany.


The recent past has brought an increasing use of integrated online
services for information discovery, the publication of personal
information and an extensive exchange of opinions. Companies and
institutions have greatly enhanced their means to collect information
about individuals at the same time, and their deployment and utilisation
is ever more sophisticated and widespread. A parallel trend exhibits the
move towards cloud computing, the outsourcing of both data and
processing to external providers.

These trends have a tremendous impact on the privacy, and terminally the
liberty of individuals users and institutions, a fact that has even
reached the legislative bodies that struggle to find regulations to
protect data while avoiding exceedingly inhibiting consequences for the
commercial affairs. The scientific advance, however, is currently
limited to highly specialised cryptographic primitives that lack general
applicability, some attempts to quantify privacy, and intermediate
solutions of outsourcing the trust to potentially independent hardware
vendors (cf. Intel SGX).

The PrInf workshop in consequence aims at uniting scientists that are
currently interested in, and working on solutions to better protect the
privacy of both the individuals with the desire to share information
online, as well as institutions that actively outsource data and
computation to cloud providers. It will hence encompass research
advances in all areas of private inference and countermeasures to
unwanted inference on personal information.

Both large-scale governmental surveillance and extensive profiling of
individuals over several integrated services has recently been enjoying
large general interest. There also have been several, rather individual
attempts at addressing these issues, but unfortunately no joint effort
covering the complete range from logics, over machine learning and data
mining, to privacy and cryptography. We hope to help build such an
inter-disciplinary forum, attracting both interest and participation of
the general audience at KI.

In particular we are interested in the following topics:
* Inference attacks on social media profiles
* Privacy metrics
* Inference and profiling prevention techniques
* Privacy preserving regression
* Privacy preserving clustering
* Privacy preserving classification
* Privacy preserving anomaly detection in social media behaviour
* Replication studies
* Data sets

Important Dates

Paper submission deadline       		June 12
Notification of acceptance			July 13
Camera Ready deadline                   	July 26
Workshop                             		Sep 21/22 (TBA)


*  Ulf Brefeld (TU Darmstadt)  
*  Thorsten Strufe (TU Dresden)

Program Committee

* Frederik Armknecht (University of Mannheim)
* Battista Biggio (University of Cagliari)
* Raphael Bost (Université de Rennes)
* Michael Brückner (Amazon)
* Christos Dimitrakakis (Chalmers)
* Krishna Gummadi (MPI for Software Systems)
* Stratis Ioannidis (Yahoo!)
* Katerina Mitrokotsa (Chalmers)
* Arvind Narayanan (Princeton)
* Yves-Alexandre de Montjoye (MIT)
* Alex Pretschner (TUM)
* Konrad Rieck (University of Göttingen)
* Björn Scheuermann (HU Berlin)

Further Information

* PrInf workshop web page:
* KI conference web page:

More information about the PET mailing list