CAREER: Dynamical Systems Modeling of Large-Scale Neural Signals Underlying Cognition

职业：认知基础大规模神经信号的动态系统建模

• 批准号: 1845836
• 负责人: Il Park
• 金额: $ 46.73万
• 依托单位: SUNY at Stony Brook
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-09-01 至 2023-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1845836&HistoricalAwards=false
• 关键词: CAREER; Dynamical; Systems; Modeling; Large

Simple mathematical expressions can often precisely describe animal behaviors during simple laboratory tasks; however, the specific inner workings of the brain reflected by its complex activity patterns have remained elusive. This project aims to transform the mathematical and computational analysis paradigm for understanding the fundamental organizational principles of neural systems through the development of novel neural data analysis tools. In particular, the computational strategies used by the animals for two behaviors: to perceive if a random visual pattern is moving toward left or right, and to combine information from visual and auditory senses to determine if they are from a potential predator, will be analyzed and visualized as hidden forces acting on the internal brain states from moment to moment. These tools will also be applied to studying disorders of consciousness in coma patients. Successful achievement of these goals will provide new analytical tools to accelerate systems-level understanding of cognition and inspire development of new treatment devices for neurological disorders. The outreach goal of this project is to increase the awareness of the general public on how the nervous system functions through short-form online educational videos that will explore misconceptions and inaccurate analogies made between computer systems and the brain, and thereby stimulate their curiosity and desire to understand how the nervous system functions at different levels. The educational goal of this project is to enrich the computational neuroscience community through curriculum development and engaging high school, undergraduate, and graduate students in the cutting-edge research and trained in advanced quantitative methodologies in computational neuroscience and neural engineering.The goals of this project are to develop both real-time and offline analysis methods to infer the normal and abnormal cognitive dynamics, that is, the internal mental processes that intelligently link sensory and motor function, manifested within the neural systems from streaming neural signals. The neural data analysis techniques will be tailored for extracting interpretable dynamics from neural time series by utilizing the low-dimensional dynamical system structure and recent developments in nonlinear state-space modeling and recurrent neural networks in machine learning. The extracted dynamics and phase portraits are expected to serve as an integral part of human understandable description of how the system operates at an intermediate scale where population state rather than individual neurons describe neural computation. The outcomes of this project will aid the development of translational medicine applications of closed-loop brain stimulation, and the education of the next generation computational neuroscientists.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.

在简单的实验室任务中，简单的数学表达式通常可以准确地描述动物的行为；然而，复杂的活动模式所反映的大脑特定的内部工作方式仍然难以捉摸。该项目旨在通过开发新的神经数据分析工具，转变数学和计算分析范式，以了解神经系统的基本组织原则。特别是，动物对两种行为所使用的计算策略：感知随机的视觉模式是向左移动还是向右移动，以及结合视觉和听觉的信息以确定它们是否来自潜在的捕食者，将被分析并可视化为不时作用于大脑内部状态的隐藏力量。这些工具也将被用于研究昏迷患者的意识障碍。这些目标的成功实现将提供新的分析工具，以加速系统层面对认知的理解，并激励神经疾病新治疗设备的开发。这个项目的推广目标是通过简短的在线教育视频提高公众对神经系统如何运作的认识，这些视频将探索计算机系统和大脑之间的误解和不准确的类比，从而激发他们了解神经系统在不同水平上如何运作的好奇心和愿望。该项目的教育目标是通过课程开发和让高中生、本科生和研究生参与尖端研究并接受计算神经科学和神经工程方面的高级量化方法培训来丰富计算神经科学界。该项目的目标是开发实时和离线分析方法来推断正常和异常的认知动力学，即通过流动神经信号在神经系统内智能地连接感觉和运动功能的内部心理过程。神经数据分析技术将利用低维动态系统结构和机器学习中非线性状态空间建模和递归神经网络的最新发展，从神经时间序列中提取可解释的动力学。提取的动力学和相图有望作为人类可理解的描述系统如何在中等尺度上运行的不可分割的一部分，其中群体状态而不是单个神经元描述神经计算。该项目的成果将有助于开发闭合式脑刺激的转化医学应用，并帮助下一代计算神经科学家的教育。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势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

Information Geometry of Orthogonal Initializations and Training

• 发表时间: 2018-10
• 期刊: ArXiv
• 作者: Piotr A. Sokól;Il-Su Park

On 1/n neural representation and robustness

• 发表时间: 2020-12
• 期刊: ArXiv
• 作者: Josue Nassar;Piotr A. Sokól;SueYeon Chung;K. Harris;Il Memming Park

Real-time variational method for learning neural trajectory and its dynamics

学习神经轨迹及其动力学的实时变分方法

• 发表时间: 2023
• 期刊: The Eleventh International Conference on Learning Representations
• 作者: Dowling, Matthew;Zhao, Yuan;Park, Il Memming
• 通讯作者: Park, Il Memming

Linear Time GPs for Inferring Latent Trajectories from Neural Spike Trains

• DOI: 10.48550/arxiv.2306.01802
• 发表时间: 2023-06
• 作者: Matthew Dowling;Yuan Zhao;Il Memming Park