A new feature set partitioning method for nearest mean classifier ensembles
Nearest Mean Classifier (NMC)provides good performance for small sample size problem. However concatenate different features into a high dimensional feature vectors and process them using a single NMC generally does not give good results because of dimensionality problem.In this new method, the fea...
保存先:
| 主要な著者: | , |
|---|---|
| フォーマット: | Conference or Workshop Item |
| 言語: | English |
| 出版事項: |
2013
|
| 主題: | |
| オンライン･アクセス: | http://repo.uum.edu.my/11967/1/PID54.pdf http://repo.uum.edu.my/11967/ http://www.icoci.cms.net.my/proceedings/2013/TOC.html |
| タグ: |
タグ追加
タグなし, このレコードへの初めてのタグを付けませんか!
|
| 要約: | Nearest Mean Classifier (NMC)provides good performance for small sample size problem. However concatenate different features into
a high dimensional feature vectors and process them using a single NMC generally does not give good results because of dimensionality problem.In this new method, the feature set is partitioned into disjoint feature subset based on diversity in ensemble.NMC ensemble is constructed by assigning each individual classifier in the ensemble with a cluster from different
feature subset.The advantage of this method is that all available information in the training set is used.There is no irrelevant feature in the training set that was eliminated.Based on experimental results the new method shows a significant improvement with high statistical confidence. |
|---|