RI: Small: Collaborative Research: A Topological Analysis of Uncertainly Representation in the Brain

描述大脑区域如何在认知中激活、协作和相互作用，使我们能够使用先进的方法来帮助人类在高压力工作中做出正确的决定，防止药物滥用，治疗神经系统疾病。本项目将从不确定性表征的角度研究认知控制，即大脑如何在不同的不确定性水平下执行相同的认知任务。基于拓扑数据分析的理论和算法，该项目将使用新颖的拓扑描述符分析脑功能MRI图像，该描述符以原则性的方式直接模拟大脑区域之间的全局相互作用。这些描述符将用于新的学习模型中，以发现对不确定性表征至关重要的大脑活动模式。该项目的成果将包括(1)不确定性表征方面的新知识，例如，与不确定性水平相关的精细尺度活动模式和大脑区域之间的相互作用；(2)新的脑成像拓扑分析工具。该项目将为计算机科学和神经科学的研究生和本科生提供研究和教育机会。这些pi还将通过纽约市立大学“现在就上大学”项目，指导来自弱势群体的学生和高中生。该项目将创建新的计算拓扑算法，从数据的内在结构中提取丰富的信息。新的机器学习方法将被创建，以便利用拓扑结构不仅进行预测，而且进行知识发现。开发一种基于拓扑特征的交互式脑成像数据探索平台。这些技术和软件将在任务诱发fMRI数据上进行验证，以产生准确性的定量评估，并表征这些方法的优点和局限性。领域专家将在验证科学假设和数据探索方面验证方法的质量。

Characterizing how brain regions activate, collaborate, and interact in cognition empowers us with advanced approaches to help humans make the right decisions on high stress jobs, prevent drug abuse, and treat neurological disorders. This project will study cognitive control in terms of the uncertainty representation, namely, how brains execute the same cognitive task with different levels of uncertainty. Based on theory and algorithms in topology data analysis, the project will analyze brain functional MRI images using novel topological descriptors, which directly model global interactions between brain regions in a principled manner. These descriptors will be used in novel learning models to discover brain activity patterns that are crucial for uncertainty representation. The outcome of the project will include (1) new knowledge in uncertainty representation, e.g., fine-scale activity patterns and interactions between brain regions correlated to the uncertainty level; (2) new topological analysis tools for brain imaging study. This project will bring research and educational opportunities to graduate and undergraduate students from both computer science and neuroscience. The PIs will also mentor students from underrepresented groups and high school students through the CUNY College Now program.This project will create new computational topology algorithms to extract rich information from the intrinsic structure of data. Novel machine learning methods will be created in order to leverage the topological structures for not only prediction, but also knowledge discovery. A novel interactive data exploration platform based on topological features will be developed for brain imaging study. These techniques and software will be validated on task-evoked fMRI data to produce quantitative assessments of accuracy and to characterize advantages and limitations of these approaches. Domain experts will validate the quality of the approach in validating scientific hypotheses and data exploration.