[Pattern recognition]. (Greece) ritis <version 4> Note 6 _ feature generation (1): Data Conversion and Dimensionality Reduction

1. Single-value decomposition: an important part of linear algebra, has been widely used in pattern recognition for dimensionality reduction and Information Retrieval Applications.

2. Independent Component Analysis

3. Non-negative matrix decomposition

4. Nonlinear Dimensionality Reduction Algorithm: ① Kernel PCA ② graph-based method (Laplace operator, local linearembedding (LLE), Isometric Mapping (Isomap ))

5. Discrete Fourier Transformation

6. discrete COs and sin Transformations

7, thehadamard Transform

8. The Haar Transform)

9. The Haar expansion Revisited)

10. Discrete Time Wavelet Transformation

11. Multi-Resolution explanation

12. Wavelet Packets)

13, a two-dimensional generation (a look at TWO-DIMENSIONALGENERALIZATIONS)

PS: It's too theoretical. I really don't have much energy for intensive reading. Let's look at a framework first!

[Pattern recognition]. (Greece) ritis <version 4> Note 6 _ feature generation (1): Data Conversion and Dimensionality Reduction