STATISTICAL APPROACHES TO UNDERSTANDING NEURAL FEATURE SELECTIVITY

理解神经特征选择性的统计方法

• 批准号: 7723146
• 负责人: Tatyana O. SHARPEE
• 金额: $ 0.05万
• 依托单位: CARNEGIE-MELLON UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2008
• 资助国家: 美国
• 起止时间: 2008-08-01 至 2009-07-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7723146
• 关键词: Affect; Auditory; Computer Retrieval of Information on Scientific Projects Database; Dimensions; Environment; Funding; Goals; Grant; Institution; Nervous system structure; Neurons; Property; Research; Research Personnel; Resources; Sensory; Simulate; Source; Statistical Methods; Stimulus; United States National Institutes of Health; Visual; neural model; object recognition; relating to nervous system; response

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. The goal of this project is to use advanced statistical methods for characterizing responses of neurons using stimuli that closely approximate those typically found in the natural sensory environment. Primarily we analyze filtering properties of visual and auditory neurons. To characterize neural responses we use a model where neural responses are affected by one or two relevant stimulus dimensions. Finding these stimulus dimensions is the most computationally intensive part of the analysis. We use a combination of simulated anneling and gradient ascent search to find the most appropriate selection of such relevant dimensions for each neuron. These results help further our understanding of how object recognition, eithr visual or auditory, is carried out in the nervous system

该子项目是利用该技术的众多研究子项目之一 资源由 NIH/NCRR 资助的中心拨款提供。子项目及 研究者 (PI) 可能已从 NIH 的另一个来源获得主要资金， 因此可以在其他 CRISP 条目中表示。列出的机构是 对于中心来说，它不一定是研究者的机构。 该项目的目标是使用先进的统计方法来表征神经元的反应，使用的刺激非常接近自然感觉环境中常见的刺激。我们主要分析视觉和听觉神经元的过滤特性。为了表征神经反应，我们使用一个模型，其中神经反应受到一个或两个相关刺激维度的影响。寻找这些刺激维度是分析中计算量最大的部分。我们结合使用模拟退火和梯度上升搜索来为每个神经元找到此类相关维度的最合适选择。这些结果有助于进一步了解物体识别（视觉或听觉）是如何在神经系统中进行的