Abstract | ||
---|---|---|
•A sample reduction strategy based on information entropy for training set is proposed. Samples are dynamically selected for training based the value of information entropy of each sample.•The information entropy is novelty defined based on the distribution of samples and their distance.•Sample reduction is introduced to support vector data description (SVDD) to speech up the training process. Meanwhile, the learning capacity of the existing SVDD is improved by introducing replace the penalty factor C with function C=1uN.•Less computing time and better classification performance are achieved by the proposed method. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1016/j.asoc.2018.02.053 | Applied Soft Computing |
Keywords | Field | DocType |
Support vector data description,Information entropy,Sample reduction,One-class classification | Kernel (linear algebra),Novelty detection,Sample (statistics),Pattern recognition,Support vector machine,Outlier,Artificial intelligence,Entropy (information theory),Mathematics,Machine learning,Computational complexity theory,Speedup | Journal |
Volume | ISSN | Citations |
71 | 1568-4946 | 1 |
PageRank | References | Authors |
0.36 | 18 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Dongdong Li | 1 | 16 | 2.91 |
Zhe Wang | 2 | 268 | 18.89 |
Chenjie Cao | 3 | 1 | 0.36 |
Yu Liu | 4 | 1 | 1.03 |