权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

CAREER: A Utility Aware Framework for Privately Sharing Individual Level Data

职业：用于私下共享个人级别数据的实用程序感知框架

基本信息

批准号：
2144684
负责人：
Liyue Fan
金额：
$ 57.49万
依托单位：
University of North Carolina at Charlotte
依托单位国家：
美国
项目类别：
Continuing Grant
财政年份：
2022
资助国家：
美国
起止时间：
2022-10-01 至 2027-09-30
项目状态：
未结题

来源：
https://www.nsf.gov/awardsearch/showAward?AWD_ID=2144684&HistoricalAwards=false
关键词：
CAREER Utility Aware Framework Privately

项目摘要

Individual level data can be collected and shared to benefit a wide range of research studies and exploratory applications. However, most privacy protecting data sharing solutions focus on well-defined aggregate analysis and do not take into account the application’s utility goals. Furthermore, current solutions may be limited in addressing the application-specific privacy needs, resulting in overly strong or inadequate privacy protection.This project develops novel privacy protecting data sharing solutions by incorporating an application’s privacy needs and utility goals in one optimization framework. The framework is compatible with a variety of rigorous privacy models. More importantly, it allows the application to customize the privacy mechanism’s structure as well as fine-grained output utility. The project shows the feasibility of customizing the framework to support real-world applications, e.g., in health and behavioral domains. Furthermore, to facilitate adoption in a wide range of domains, the project estimates the utility loss via statistical interactions in the data and develops computationally efficient techniques to solve large scale problems. Moreover, the project studies domain-specific privacy risks, e.g., in rapidly growing applications, to formulate the evolving privacy needs in the framework. The results of the project will benefit research studies and applications in a variety of domains that rely on individually contributed data, such as, behavioral and health studies. The project also integrates research and education with a number of interconnected activities, including curriculum development, student mentoring, interdisciplinary collaboration, and K-12 outreach.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

个人层面的数据可以收集和共享，以利于广泛的研究和探索性应用。然而，大多数隐私保护数据共享解决方案都集中在定义良好的聚合分析上，而没有考虑应用程序的实用目标。此外，目前的解决方案可能局限于解决特定应用程序的隐私需求，导致过于强大或不足的隐私保护。本项目通过将应用程序的隐私需求和实用程序的目标结合在一个优化框架中，开发新的隐私保护数据共享解决方案。该框架与各种严格的隐私模型兼容。更重要的是，它允许应用程序自定义隐私机制的结构以及细粒度的输出实用程序。该项目显示了定制框架以支持真实世界应用程序的可行性，例如，在健康和行为领域。此外，为了便于在广泛的领域中采用，该项目通过数据中的统计交互来估计效用损失，并开发计算效率高的技术来解决大规模问题。此外，该项目还研究了特定领域的隐私风险，例如，在快速增长的应用程序中，在框架中制定不断变化的隐私需求。该项目的结果将有利于依赖个人贡献数据的各种领域的研究和应用，例如行为和健康研究。该项目还将研究和教育与一些相互关联的活动相结合，包括课程开发，学生辅导，跨学科合作和K-12外展。该奖项反映了NSF的法定使命，并被认为值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估来支持。

项目成果

期刊论文数量（4）

专著数量（0）

科研奖励数量（0）

会议论文数量（0）

专利数量（0）

Mitigating Membership Inference in Deep Survival Analyses with Differential Privacy

DOI：
10.1109/ichi57859.2023.00022
发表时间：
2023-06
期刊：
2023 IEEE 11th International Conference on Healthcare Informatics (ICHI)
影响因子：
0
作者：
Liyue Fan;Luca Bonomi
通讯作者：
Liyue Fan;Luca Bonomi

Enabling Health Data Sharing with Fine-Grained Privacy

DOI：
10.1145/3583780.3614864
发表时间：
2023-10
期刊：
Proceedings of the 32nd ACM International Conference on Information and Knowledge Management
影响因子：
0
作者：
Luca Bonomi;Sepand Gousheh;Liyue Fan
通讯作者：
Luca Bonomi;Sepand Gousheh;Liyue Fan

Private Data Synthesis from Decentralized Non-IID Data