F-008 An Novel Soft Margin Classifier Using Genetic Algorithm
スポンサーリンク
概要
- 論文の詳細を見る
In many high dimensional classification problems, though the different class data are not linearly separable, we prefer a linear classifier over an over-trained high variance boundary. By limiting to linear boundary, in many practical applications, we can ignore noisy samples. Linear support vector machine is such a Soft margin classifier. In this work, we propose a genetic algorithm approach to construct the soft margin classifier. We simulated the algorithm and did several experiments with synthetic and real life data. The results are evaluated and found to be very near to optimum. The quality of the results and computation costs are compared with existing algorithms.
- FIT(電子情報通信学会・情報処理学会)推進委員会の論文
- 2008-08-20
著者
関連論文
- 6Q-6 自己組織化マップを用いた動径基底関数ネットワークの構築(ニューラルネット(2),学生セッション,人工知能と認知科学)
- F-029 自己組織化マップを用いた動径基底関数ネットワークの性能評価(F分野:人工知能・ゲーム,一般論文)
- F-008 An Novel Soft Margin Classifier Using Genetic Algorithm
- F-007 An Efficient Construction of RBF Network Based on Training by SOM
- G-010 遺伝子発現データを用いた病理診断における遺伝子選択手法の提案(生体情報科学,一般論文)
- H_027 A Health Life-style Checke List and Analysis