Sprint 1 Webinar scheduled for Oct 14th @ 11:30 EDT

Calling all participants! The organizers of the Differential Privacy Temporal Map Challenge (DeID2) will be hosting an informational webinar this coming Wednesday Oct 14th from 11:30 am - 12:30 pm EDT. This is a great chance to learn more about differential privacy and how to participate in Sprint 1.

Register here: https://attendee.gotowebinar.com/register/7739330816637578765

After registering, you will receive a confirmation email containing information about joining the webinar. The webinar will also be recorded, and we will post the recording after the session is complete.

By now you’ve probably had a chance to look around the website some, check out the competitor’s pack, and think a little bit about the problem. We hope you like it-- it’s a fundamental problem underlying many practical applications, and we’ve selected a real world data set with real world properties like sparseness.

But we didn’t just select this problem for its applications, we wanted to make sure the problem was designed to be amenable to many different strategies: Have you noticed that it could also be a clustering problem? A signal processing problem? A social network problem? An individual behavior modeling problem?

There’s no obligation to look at the problem only as a pile of event records with a high sensitivity. We’ve also provided simulated individuals. To generate these individuals, we used fine-grained geographic information from the original source data, in order to ensure meaningful correlations between the event types associated with a given simulated individual.

You can look at this problem by neighborhoods, by patterns across time, by relationships between event types, as a set of individuals, or any combination of the above. In the webinar tomorrow, we’ll walk through the challenge structure, the competitor’s pack… and we’ll highlight some points of interest about the problem itself.

Successful privacy algorithms tend to find efficient ways to capture the fundamental patterns underlying human data. Happy adventures in space time!

Thanks everyone for the great session yesterday! ICYMI here are the recorded webinar and slides.