An approach to incremental SVM learning algorithm. Xiao, R., Wang, J., & Zhang, F. In Tools with Artificial Intelligence, 2000. ICTAI 2000. Proceedings. 12th IEEE International Conference on, pages 268--273, 2000.
abstract   bibtex   
The Classijication algorithm based on Support Vector Machine (SVM) now attracts more attentions due to its pelfect theoretical properties and good empirical results. In this papel; we first analyze the properties of SV set thoroughly, then introduce a new learning method, which extends the SVM Classijication algorithm to incremental learning area. The theoretical bases of this algorithm are the classijication equivalence of the SV set and the training set. In this algorithm, the knowledge is accumulated in the process of incremental Learning. In addition, unimportant samples are discarded optimally by LRU scheme. The theoretical analysis and experimental results show that this algorithm could not only speedup the training process, but also reduce the storage cost, while the classijkation precision is also guaranteed.
@InProceedings{Xiao2000,
  Title                    = {An approach to incremental SVM learning algorithm},
  Author                   = {Xiao, Rong and Wang, Jicheng and Zhang, Fayan},
  Booktitle                = {Tools with Artificial Intelligence, 2000. ICTAI 2000. Proceedings. 12th IEEE International Conference on},
  Year                     = {2000},
  Pages                    = {268--273},

  Abstract                 = {The Classijication algorithm based on Support Vector 
Machine (SVM) now attracts more attentions due to its 
pelfect theoretical properties and good empirical results. 
In this papel; we first analyze the properties of SV set 
thoroughly, then introduce a new learning method, which 
extends the SVM Classijication algorithm to incremental 
learning area. The theoretical bases of this algorithm are 
the classijication equivalence of the SV set and the 
training set. In this algorithm, the knowledge is 
accumulated in the process of incremental Learning. In 
addition, unimportant samples are discarded optimally by 
LRU scheme. The theoretical analysis and experimental 
results show that this algorithm could not only speedup 
the training process, but also reduce the storage cost, 
while the classijkation precision is also guaranteed.},
  Timestamp                = {2014.10.24}
}

Downloads: 0