project // research // linear algebra
Structural Dependencies of Handwritten Digits (MNIST)_
A UofT research project that applies PCA to analyze geometric structure fragments of MNIST digits, revealing linear dependencies and interpretable subspaces in the data.
多伦多大学研究项目：使用 PCA 对 MNIST 手写数字的几何结构片段进行分析， 揭示数据中的线性依赖关系与可解释的低维子空间结构。
Python
NumPy
Scikit-learn
PCA
Linear Algebra
Output
Full Essay
methods / results / discussion
Approach
pca // geometry // dependencies
- • Constructed feature representations that emphasize structural fragments of digits.
- • Applied PCA to identify dominant axes of variation and interpretable linear subspaces.
- • Analyzed linear dependencies to connect geometry with dimensionality reduction outcomes.
- • 构造强调“结构片段”的特征表示，使几何形态更易分析。
- • 使用 PCA 提取主变化方向与可解释的低维子空间。
- • 分析线性依赖关系，将几何结构与降维结果建立联系。
Tech Stack
core tools
Python / NumPy
matrix ops
Scikit-learn
pca implementation
Linear Algebra
structure & dependencies
Primary Link
essay