Kernel MethodsIt is a type of pattern recognition algorithm whose purpose is to find and learn the mutual relationships in a set of data.

The kernel method is based on the following assumption: "A point set that cannot be linearly separated in a low-dimensional space may become linearly separable after being transformed into a point set in a high-dimensional space."

Basic understanding of kernel methods: Patterns in the original data can be simply expressed in a certain dimensional space, where "simple expression" refers to linear relationships. Kernel methods solve the problem of how to map data to a specific dimensional space.

The general processing steps of the kernel method are:

• Process various types of data into corresponding kernel matrices;
• Use the kernel matrix to find patterns in the data.

From another perspective, the kernel method can also be understood as a mathematical construction of a more interpretable single-layer neural network.

Currently, kernel methods have been widely used in text, speech, image and other fields.

References

【1】http://yongqwang.com/public/res_mat/K_Intro_01.pdf