We invite participation in the workshop on “Learning and privacy with incomplete data
and weak supervision” as a part of the NIPS 2015 conference. We welcome and encourage
multidisciplinary submissions on the topic of learning individual traits from
aggregate and noisi(fied) data, and the protection against attacks of this sort.
novel algorithms and theoretical analysis on “weakly-supervised” learning, e.g.
multiple instance learning, learning from label proportions, noisy labels, ecological inference
field-specific applications in computer, biomedical and social sciences
innovative methods and perspectives on data collection, aggregation and obfuscation
privacy aspects of methods for data sharing and release
application demos of the above (to run live at the workshop)
social and legal implications and open problems
Submissions must be 4-8 pages long (+1 for references), and adhere to the NIPS format.
Papers submission is here. Please leave authors information visible.
With regard to the demos, we encourage the use of publicly accessible
datasets and code in the interest of reproducibility; this will be valued positively
A subset of the accepted papers will be presented orally; all accepted papers will
be part of a poster/demo session. Also, depending on quality and number of submissions, and subject to the organizers’ judgement, a selection of the work presented at the workshop will be part of
a special issue on the Journal of Privacy and Confidentiality,
NEW: The organizers will nominate the authors of the top quality submitted work for Best Paper Award, sponsored by Google.
NEW: We DO accept abstract submissions of work recently published or currently under review.
Submission deadline for papers and demos: October 10 17 2015 (extended)
Notification of acceptance: October 27 30 2015 (extended)
Notification of acceptance on the Journal of Privacy and Confidentiality: TBA