Support Vector Machine Classification

The machine in the classifier is actually an algorithm that is extended to machine learning. An algorithm is obtained by training samples and calculating the attributes and features of samples.

In a narrow sense, the classification methods can be classified into Support Vector Machines and unsupported vector machines. Support Vector Machines are the severability of the samples to be classified in the vector space. The Support Vector has the following properties: The Support Vector within the 1 boundary must be in the correct partition on the Self-interval boundary; 2 the Support Vector does not appear in the correct partition at the interval. 3. The non-SVM must be in the correct partition with intervals.

For non-linear division, the Support Vector Machine classification algorithm introduces the maximum interval principle and kernel technique. The kernel technique converts the classification problem of the original space that requires the use of noisy Curved Surface Differentiation into the problem of the use of hyperplane division, converts a non-linear space into a linear space to reduce the difficulty of classification.

Multiclass classification: Most SVM classifications tend to be classified into two categories. For the classification of remote sensing images, the classification of the two categories is rare, in more cases, we are faced with multiclass classification issues. For multiclass classification machines, the general idea is to construct a series of two types of classification machines, each of which divides one of them into the other types, and then deduce the ownership of an input x.