Dependable Data Driven Discovery

可靠的数据驱动发现

• 批准号: 2152117
• 负责人: Wallapak Tavanapong
• 金额: $ 299.9万
• 依托单位: Iowa State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-07-01 至 2027-06-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2152117&HistoricalAwards=false
• 关键词: Dependable; Data; Driven; Discovery

Data-driven decisions are becoming increasingly critical for the well-being of individuals and society; and the nation and the world’s reliance on such decisions is likely to increase tremendously in the next decade. However, data science lifecycles are assembled and operated by a wide variety of individuals and institutions with varying levels of expertise. While to err is human, the consequences of errors in critical data science lifecycles can be catastrophic. Unreliable decisions can potentially have enormous negative impacts such as loss of life, widespread contagions, and economic depression. To respond to this critical challenge, this NSF Research Traineeship project is establishing a graduate program of study, "Dependable Data Driven Discovery (D4)" that engages faculty across multiple disciplines to train students with diverse scientific backgrounds to recognize risks to dependable data driven discovery and to develop corresponding mitigation strategies. One hundred students are expected to participate from disciplines such as computer science, mathematics, statistics, bioengineering, and computational biology, including 32 funded graduate (MS & PhD) trainees and 16 undergraduate students from minority groups who are underrepresented in their participation in these fields. Trainees will gain a holistic perspective of the entire data science lifecycle through coursework as well as collaborative, transdisciplinary research. Three focal areas comprise the D4 research and training agenda. First is a focus on formal foundations, methodology, and tools for a dependable data driven discovery framework. Second is an examination of risk mitigation methods to handle noise in data, limited training data, and uncertainty prediction and interpretability of machine learning models. The third focal area addresses quality assurance issues for protein function prediction and cellular engineering processes to direct undifferentiated cells into mature, functional cells with dependable data science lifecycles. The project will develop a new graduate certificate in dependable data science to train students in dependability issues within data science lifecycles. Through coursework, trainees will experience the entire data science lifecycle several times, each with an increasingly deeper understanding of the risks, measures, and risks mitigation mechanisms. Trainees will also engage with industry partners to examine their data science lifecycles and discuss risk mitigation methods. The capstone project course will reinforce trainees’ technical skills to address the above research problems in data science, biological science, and engineering as well as written and oral communication skills. Trainees will interact with outside collaborators through the D4 seminars, gain experiential learning with industry partners, conduct original research through capstone projects, and work in internships, resulting in awareness of dependable data science lifecycles and risk mitigation mechanisms across Iowa State University, local industry, NGOs, and government. The project will engage in outreach activities through two existing Iowa State University infrastructures: ISU Science Bound and ISU Extension and Outreach with Iowa 4-H.The NSF Research Traineeship (NRT) Program is designed to encourage the development and implementation of bold, new potentially transformative models for STEM graduate education training. The program is dedicated to effective training of STEM graduate students in high priority interdisciplinary or convergent research areas through comprehensive traineeship models that are innovative, evidence-based, and aligned with changing workforce and research needs.This project is jointly funded by NRT and the Established Program to Stimulate Competitive Research (EPSCoR).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.

数据驱动的决策对个人和社会的福祉越来越重要;在未来十年，国家和世界对此类决策的依赖可能会大大增加。然而，数据科学生命周期是由具有不同专业知识水平的各种个人和机构组装和操作的。虽然犯错是人的本性，但关键数据科学生命周期中的错误后果可能是灾难性的。不可靠的决策可能会产生巨大的负面影响，如生命损失、广泛的传染和经济萧条。为了应对这一关键挑战，NSF研究培训项目正在建立一个研究生课程，“依赖数据驱动的发现（D4）”，该课程涉及多个学科的教师，以培训具有不同科学背景的学生认识到可靠数据驱动发现的风险，并制定相应的缓解策略。预计将有100名来自计算机科学，数学，统计学，生物工程和计算生物学等学科的学生参加，其中包括32名受资助的研究生（MS PhD）学员和16名来自少数群体的本科生，他们在这些领域的参与人数不足。学员将通过课程作业以及跨学科的协作研究获得整个数据科学生命周期的整体视角。D4研究和培训议程包括三个重点领域。首先是关注可靠的数据驱动发现框架的正式基础、方法和工具。第二是检查风险缓解方法，以处理数据中的噪声，有限的训练数据以及机器学习模型的不确定性预测和可解释性。第三个重点领域涉及蛋白质功能预测和细胞工程过程的质量保证问题，以指导未分化细胞进入成熟的功能细胞，并具有可靠的数据科学生命周期。该项目将开发一个新的可靠数据科学研究生证书，以培训学生在数据科学生命周期内的可靠性问题。通过课程，学员将多次体验整个数据科学生命周期，每次都对风险，措施和风险缓解机制有更深入的了解。学员还将与行业合作伙伴合作，检查他们的数据科学生命周期，并讨论风险缓解方法。顶点项目课程将加强学员的技术技能，以解决上述数据科学，生物科学和工程方面的研究问题，以及书面和口头沟通技巧。学员将通过D4研讨会与外部合作者互动，与行业合作伙伴进行体验式学习，通过顶点项目进行原创研究，并在实习中工作，从而提高对爱荷华州州立大学，当地行业，非政府组织和政府的可靠数据科学生命周期和风险缓解机制的认识。该项目将通过两个现有的爱荷华州州立大学的基础设施参与外展活动：ISU科学约束和ISU扩展和外展与爱荷华州4-H. NSF研究培训（NRT）计划旨在鼓励开发和实施大胆的，新的潜在变革模型STEM研究生教育培训。该计划致力于通过创新，循证，并与不断变化的劳动力和研究需求保持一致。该项目由NRT和刺激竞争性研究既定计划（EPSCoR）联合资助该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Mutation-based Fault Localization of Deep Neural Networks

• DOI: 10.1109/ase56229.2023.00171
• 发表时间: 2023-09
• 期刊: 2023 38th IEEE/ACM International Conference on Automated Software Engineering (ASE)
• 作者: Ali Ghanbari;Deepak-George Thomas;Muhammad Arbab Arshad;Hridesh Rajan

What kinds of contracts do ML APIs need?

• DOI: 10.1007/s10664-023-10320-z
• 发表时间: 2023-07
• 期刊: Empirical Software Engineering
• 影响因子: 4.1
• 作者: S. K. Samantha;Shibbir Ahmed;S. Imtiaz;Hridesh Rajan;G. Leavens

Fix Fairness, Don’t Ruin Accuracy: Performance Aware Fairness Repair using AutoML

• DOI: 10.1145/3611643.3616257
• 发表时间: 2023-06
• 期刊: Proceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering
• 作者: Giang Nguyen-;Sumon Biswas;Hridesh Rajan

Towards Understanding Fairness and its Composition in Ensemble Machine Learning

• DOI: 10.1109/icse48619.2023.00133
• 发表时间: 2022-12
• 期刊: 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE)
• 作者: Usman Gohar;Sumon Biswas;Hridesh Rajan

Design by Contract for Deep Learning APIs

• DOI: 10.1145/3611643.3616247
• 发表时间: 2023-11
• 期刊: Proceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering
• 作者: Shibbir Ahmed;S. Imtiaz;S. K. Samantha;Breno Dantas Cruz;Hridesh Rajan