複数評価関数を考慮した遺伝的アルゴリズムによるニューラルネットワークの荷重値決定法

• 渡邉 伸司

  岐阜工業高等専門学校専攻科(電子システム工学専攻)

• 蔡 篤儀

  岐阜工業高等専門学校電気工学科

• 小島 克之

  常葉学園浜松大学経営情報学部経営情報学科

• 山田 功

  岐阜医療技術短期大学診療放射線技術学科

書誌事項

タイトル別名

• Determination of Weighting Coefficients of Neural Networks Trained by Multiple-Fitness-Based Genetic Algorithms

抄録

In our previous study we reported an artificial neural network (NN) trained by genetic algorithms (GA) instead of that trained by the conventional backpropagation (BP) technique, and applied this GA training method to classifying heart diseases in echocardiograms. The results showed that the GA-based NN was superior to the BP-based NN in terms of classification rate as well as the ease of numerical computation. However, the reliability issue was the drawback of the proposed method. In order to overcome the shortcoming, in the present study we propose an improved method by employing multiple fitness functions in the selection operation used in the GA. Our preliminary results reveal the effectiveness of the newly proposed method.

収録刊行物

• Medical Imaging and Information Sciences

  Medical Imaging and Information Sciences 15 (1), 53-60, 1998

  Medical Imaging and Information Sciences

キーワード

• computer - aided diagnosis
• genetic algorithm
• neural network
• echocardiography

詳細情報

• CRID: 1390001205456392448
• NII論文ID: 130004647607
• DOI: 10.11318/mii1984.15.53
• ISSN: 09101543
• 本文言語コード: ja
• データソース種別:
  • JaLC
  • CiNii Articles
• 抄録ライセンスフラグ: 使用不可