An ISVM Algorithm Based on High-Dimensional Distance and Forgetting Characteristics

Algorithm 1

The process of HDFC-ISVM algorithm.

	Premise: suppose , T is the initial sample set, and is the new sample set;
	Objective: it will find the SVM classifier based on .
	Step 1. The initial dataset is trained to obtain the initial classifier and the initial SV set , and the forgetting factor corresponding to each sample in the set is calculated.
	Step 2. Check whether the incremental set exists. If not, the algorithm ends, and is the final classifier; otherwise, it will enter into step 3.
	Step 3. For the incremental set , the forgetting factor of every sample of is calculated according to the classifier .
	Step 4. Set , it will select the samples whose forgetting factor satisfied to construct the set .
	Step 5. A new round of SVM training is carried out for the dataset to obtain a new classifier .
	Step 6. For the classifier , the above threshold adjustment rule is used to update the forgetting factor of each sample in the dataset , setting and then turning to step 2.