Research Initiation Award: Thermal Decomposition of Four-membered Heterocyclic Peroxides, Data Mining in Nonadiabatic Trajectories, and Chemiexcitation Efficiency

HBCU-UP的研究启动奖为STEM教师开展研究活动提供支持，以进一步提高他们的研究能力和效率，并加强hbcu的STEM研究和本科教育。该奖项授予杰克逊州立大学，有潜力改善生物图像，培养机器学习和人工智能方面的本科生，并为STEM课程生产先进技术的教育模块。该项目旨在利用机器学习的新应用来识别光发射器中哪些原子主导其化学激发效率，以改进生物成像设备。该项目的意义在于推进热分解机器学习的知识，并协助合成新一代发光体，提高生物发光成像的图像质量，改善肿瘤治疗的治疗效果。学员将接受关键工作技能的培训。本项目旨在使用贝叶斯支持向量机（BSVM）学习模型来揭示合成光源中高化学激发效率的起源。该方案的新颖之处在于：(1)通过优化参考轨迹周围的CI接缝得到的圆锥相交（CI）拓扑，阐述了轨迹的激励；2)通过BSVM分离矩阵投影轨迹在任意时刻的能量状态；3)搜索描述决策边界的最具影响力的特征，对表面跳跃轨迹进行统计分类；4)刻画离分离矩阵最远的特征，高概率地预测能量态；5)建立了初始相空间坐标与过渡阈值之间的关系，便于高效新型光源的设计；6)开发多个小型计算模块，每个模块将作为HBCU本科生的研究课题。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。

HBCU-UP’s Research Initiation Awards provide support for STEM faculty to pursue research activities to further their research capabilities and effectiveness and enhance STEM research and undergraduate education at HBCUs. This award to Jackson State University has the potential to improve bio-imagery, train undergraduate students in machine learning and artificial intelligence, and produce education modules in advanced technology for STEM courses. The project aims to use a novel application of machine learning to discern what atoms in a light emitter dominate its chemiexcitation efficiency to improve bio-imaging devices. The significance of this projects is in advancing knowledge of machine learning of thermolysis and assisting in the synthesis of a new generation of light emitters that advances image quality for bioluminescent imaging and improved treatment outcomes for tumor therapy. Student participants will receive training in critical workforce skills. This project seeks to use the Bayesian Support Vector Machine (BSVM) learning model to uncover the origins of high chemiexcitation efficiency in synthesized light emitters. The project is novel in that it (1) expounds the excitation of the trajectories via the conical intersection (CI) topology obtained by optimizing the CI seams around the reference trajectories; 2) projects the energy state of a trajectory at any instant via the BSVM separatrix; 3) searches the most influential features that describe the decision boundary and classifies surface hopping trajectories statistically; 4) characterizes the features that are farthest from the separatrix and predicts the energy states with high probability; 5) establishes the relation between initial phase space coordinates and transition thresholds to facilitate the designing process for new light emitters with high lighting efficiency; and 6) develops multiple small computational modules, each of which will be used as research topics for HBCU undergraduate student research.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.