CAREER: HayaRupu: Accelerating Natural Hazard Engineering with AI-Driven Discovery Loops

职业：HayaRupu：利用人工智能驱动的发现循环加速自然灾害工程

• 批准号: 2339678
• 负责人: Krishna Kumar
• 金额: $ 50万
• 依托单位: University of Texas at Austin
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2024
• 资助国家: 美国
• 起止时间: 2024-08-01 至 2029-07-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2339678&HistoricalAwards=false
• 关键词: CAREER; HayaRupu; Accelerating; Natural; Hazard

The HayaRupu project aims to accelerate advancements in engineering and scientific research through artificial intelligence (AI), utilizing landslide hazards as a key area for demonstration. Central to this initiative is the AI-accelerated scientific discovery loop, a framework that integrates AI into every stage of scientific inquiry, from knowledge discovery to hypothesis testing and modeling. This approach leverages machine learning (ML) for sophisticated pattern recognition and anomaly detection, enabling the extraction of new insights from complex datasets. A significant focus of HayaRupu is identifying and addressing knowledge gaps in natural hazard engineering by building a context-aware knowledge graph. Furthermore, the project innovates by integrating AI in numerical simulations, exploiting the speed of AI and the accuracy of numerical simulations to develop novel optimization strategies. The physics-aware AI methods bridge the gap between simulated environments and real-world applications. HayaRupu's approach exemplifies how physics-aware AI can accelerate scientific progress. A key aspect of HayaRupu is its educational outreach, which involves creating new AI-assisted scalable and personalized learning for engineering students. This educational initiative supports the development of future engineers with skills in cutting-edge AI technologies, enhancing diversity in STEM fields and contributing to a skilled workforce adept in integrating AI and natural hazard engineering.The HayaRupu framework (Japanese for FastLoop) will accelerate discoveries in natural hazards engineering (NHE) by applying Artificial Intelligence (AI) to facilitate knowledge discovery and accelerate NextGen AI-embedded simulation tools for exascale simulations through physics-aware AI techniques. The work provides novel AI solutions to accelerate discoveries in natural hazard engineering. It has three key intellectual merits: (i) context-aware knowledge graphs as reasoning engines to enable new data-driven discoveries and derive fundamental multi-scale equations through geometric deep learning, (ii) building the NextGen AI-accelerated differentiable simulators offering a new paradigm for solving inverse and design problems, (iii) creating an integrated framework of Large Language Model-enabled robust end-to-end automated workflow design to demonstrate the potential of AI in driving scientific advances. HayaRupu is developing a personalized and scalable AI tutor to transform engineering education by offering future engineers a personalized learning environment, quizzes, and support. Key educational outreach initiatives include organizing Tween Code Clubs at the Austin Public Library to promote computational and AI literacy among young learners and offering targeted programs in low-income and underrepresented communities. Additionally, the project collaborates with the Code@TACC program to inspire high school students, especially those from marginalized backgrounds, towards STEM careers. Through these efforts, HayaRupu advances scientific understanding and fosters a diverse and inclusive environment in science and technology education.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.

HayaRupu项目旨在通过人工智能（AI）加速工程和科学研究的进步，利用滑坡灾害作为示范的关键领域。该计划的核心是人工智能加速的科学发现循环，该框架将人工智能集成到科学探究的每个阶段，从知识发现到假设检验和建模。这种方法利用机器学习 (ML) 进行复杂的模式识别和异常检测，从而能够从复杂的数据集中提取新的见解。 HayaRupu 的一个重要重点是通过构建上下文感知知识图来识别和解决自然灾害工程中的知识差距。此外，该项目还通过将人工智能集成到数值模拟中进行创新，利用人工智能的速度和数值模拟的准确性来开发新颖的优化策略。物理感知人工智能方法弥合了模拟环境和现实应用之间的差距。 HayaRupu 的方法体现了物理感知人工智能如何加速科学进步。 HayaRupu 的一个关键方面是其教育推广，其中包括为工程专业学生创建新的人工智能辅助的可扩展和个性化学习。这项教育计划支持培养拥有尖端人工智能技术技能的未来工程师，增强 STEM 领域的多样性，并为擅长整合人工智能和自然灾害工程的熟练劳动力做出贡献。HayaRupu 框架（日语为 FastLoop）将加速以下领域的发现：通过应用人工智能 (AI) 来促进知识发现并加速下一代 AI 嵌入式模拟工具通过物理感知 AI 技术进行百亿亿次模拟，从而实现自然灾害工程 (NHE)。这项工作提供了新颖的人工智能解决方案，以加速自然灾害工程的发现。它具有三个关键的智力优点：（i）上下文感知知识图作为推理引擎，以实现新的数据驱动发现并通过几何深度学习推导出基本的多尺度方程，（ii）构建下一代人工智能加速的可微分模拟器，提供解决逆向和设计问题的新范式，(iii) 创建一个支持大型语言模型的强大的端到端自动化工作流程设计的集成框架，以展示人工智能在推动科学进步方面的潜力。 HayaRupu 正在开发个性化且可扩展的人工智能导师，通过为未来的工程师提供个性化的学习环境、测验和支持来改变工程教育。主要的教育推广举措包括在奥斯汀公共图书馆组织吐温代码俱乐部，以提高年轻学习者的计算和人工智能素养，并为低收入和代表性不足的社区提供有针对性的项目。此外，该项目还与 Code@TACC 计划合作，激励高中生，特别是来自边缘化背景的高中生，走向 STEM 职业。通过这些努力，HayaRupu 促进了科学理解，并在科技教育中营造了多元化和包容性的环境。该奖项反映了 NSF 的法定使命，并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，被认为值得支持。