Dimensionality reduction and intrinsic dimension.

降维和内在维度。

• 批准号: 2431450
• 金额: --
• 依托单位: University of Birmingham
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2020
• 资助国家: 英国
• 起止时间: 2020 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-2431450
• 关键词: Dimensionality; reduction; intrinsic; dimension.

This proposal focuses on the intrinsic dimension of datasets. We propose new ways of estimating it by taking into account the problem at-hand. We use several already proposed approaches of estimation but customize them based on our intuition and other physical constraints associated with the field or sector we deal with. We will also compare our results with other existing methods to see whether our new approaches yield better approximations for the intrinsic dimension.

该提案侧重于数据集的内在维度。我们提出了新的方法来估计它考虑到手头的问题。我们使用几种已经提出的估计方法，但根据我们的直觉和与我们处理的领域或部门相关的其他物理约束来定制它们。我们还将比较我们的结果与其他现有的方法，看看我们的新方法是否产生更好的近似的内在尺寸。