In today’s digital world, nearly every action we take generates data. These data sets, when analyzed, can reveal valuable insights to organizations and companies both in and outside of the Public Safety Sector. However, privacy concerns currently limit how and when these datasets can be analyzed and since current data anonymization algorithms are mostly inadequate, leveraging this data in a meaningful way becomes difficult. Differential privacy techniques provably prevent the disclosure of personally identifiable information while preserving the utility of the dataset for analysis. This challenge was designed to find ways to accelerate the pace at which progress is being made on these techniques so valuable data can be used in a variety of ways.
NIST PSCR collaborated with HeroX, NASA Tournament Labs, and Topcoder to design and implement a crowdsourcing challenge to advance the field of differential privacy. This challenge was the first of its kind to benchmark the effectiveness of different solutions against each other. The multi-phased challenge broke the problem down into smaller, more digestible pieces and allowed for iterations between stages to maximize the project’s effectiveness and success. This approach simultaneously made the problem easier to address and provided more opportunities for head to head competition, which has proven key to maintaining forward momentum. Additionally, outreach efforts for this challenge brought in innovators from outside the privacy community. These new participants help raise the profile of the problem while bringing in more diverse perspectives.
The challenge received 132 entries and selected 14 winners to share the $150,000 USD prize purse. The results of this challenge surpassed all expectations of NIST PSCR, and they intend to share the challenge results in upcoming publications. The source code from several of the winning teams was open-sourced, and it is publicly posted. NIST made significant progress in this field, but there is still further research needed. NIST is considering future challenges to continue pushing forward with the differential privacy efforts.
Grand Prize Winners:
- 1st Place: rmckenna - 902,307 ($25,000)
- 2nd Place: ninghui - 870,097 ($15,000)
- 3rd Place: privbayes - 823,512 ($10,000)
- 4th Place: gardn999 - 768,802 ($5,000)
- 5th PLace: manisrivastava - 541,494 ($3,000)