CREST HBCU-RISE: Advancing Theoretical Artificial Intelligence Infrastructure for Modern Data Science Challenges

在科学技术卓越研究中心（CREST HBCU- rise）的支持下，该项目旨在改善田纳西州立大学（TSU）的理论人工智能（AI）基础设施，以应对重要的数据科学挑战。虽然人工智能应用激增，但在构建适用于安全关键应用的可靠和安全的人工智能系统所需的基本数学理论方面仍然存在差距。这个项目试图解决这个问题。预期成果包括人工智能理论的科学知识进步及其在网络安全、生物信息学和农业方面的实际应用。此外，该项目旨在加强跨学科合作，促进STEM领域更广泛的参与，增强托大的研究竞争力。总体目标是提高新兴人工智能领域的研究能力，并与托国立的三个学院（工程、生命与物理科学、农业）合作，开发一种全面的方法来教育和培养博士生。该研究将为精确和强大的机器学习开发数学理论和实用算法，可用于推进隐私保护人工智能，增强冷冻电子显微镜成像的蛋白质结构建模以及精准农业的最佳特征选择的研究。将研究用于神经网络训练的高维流形几何，用于私有人工智能同态加密的网络线性化，用于冷冻电子显微镜的子空间分割和机器学习的系统集成，以及用于作物参数和产量估计的多尺度传感数据的质量评估。该项目在三所学院实施了连贯的人工智能教育课程，包括开发教材，为学生组织专业发展活动，为博士生提供指导，以及采购研究设备以支持这项研究。CREST HBCU-RISE计划支持hbcu机构研究能力的扩展以及成功培训STEM博士生。该奖项反映了美国国家科学基金会的法定使命，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。

With support from the Centers of Research Excellence in Science and Technology HBCU Research Infrastructure for Science and Engineering (CREST HBCU-RISE), this project aims to improve the theoretical Artificial Intelligence (AI) infrastructure at Tennessee State University (TSU) for addressing important data science challenges. While AI applications have proliferated, there remains a gap in the fundamental mathematical theories required to construct reliable and secure AI systems suitable for safety-critical applications. This project seeks to address this. The expected outcomes include the advancement of scientific knowledge in AI theory and its practical applications in cybersecurity, bioinformatics, and agriculture. Additionally, this project aims to enhance interdisciplinary collaboration, promote broader participation in STEM fields, and strengthen research competitiveness at TSU.The overarching goal is to advance research capabilities in emerging AI areas and develop a comprehensive approach to educate and train Ph.D. students in collaboration with three TSU colleges: Engineering, Life and Physical Sciences, and Agriculture. The research will develop mathematical theory and practical algorithms for accurate and robust machine learning that can be applied for advancing research in privacy-preserving AI, protein structure modeling with enhanced cryo-electron microscopy imaging, and optimal feature selection for precision agriculture. High-dimensional manifold geometries for neural networks training, network linearization for homomorphic encryption in private AI, systematic integration of subspace segmentation and machine learning for cryo-electron microscopy, and quality assessment of multi-scale sensing data for crop parameters and yield estimation will be investigated. This project implements a coherent curriculum across three colleges for AI education, which includes developing educational materials, organizing professional development activities for students, providing Ph.D. student mentoring, and procuring research equipment to support this research. The CREST HBCU-RISE program supports the expansion of institutional research capacity as well as the successful training of doctoral students in STEM at HBCUs.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.