영문내용
(English Abstract) |
Recently, various types of neural network modes have been used successfully to applications in pattern recognition, control, signal processing, and so on. However, the previous models are not suitable for hardware implementation due to their complexity.
In this paper, we present a survey of stochastic analysis for Langevine Comepetitive Learning Algorithm and show an experimental results of pattern recognition for multispectral earth remote sensing data. The Langevine comepetitive learning algorithm uses a time-invariant learning rate whereas the conventional competitive learning developed by Kohonen and others uses a time-varing learning rate. In the Langevine comepetitive learning, there is a binary reinforcement function in order to compensate for the lowered learning ability due to the constant learning rate.
The experimental results conducted with two different types of data indicates the superiority of the proposed method in comparison to the original comepetitive learning. |
