International Symposium on Nonlinear Theory and its Applications
Cellular Neural Networks with Hopfield Neural Networks Considering the Confidence Degree
Yasuhiro Ueda, Masakazu Kawahara, Yoko Uwate, Yoshifumi Nishio,
PDF download (98KB)
In this study, we propose cellular neural networks (CNN) with Hopfield neural networks (Hopfield NN) considering the confidence degree. The Hopfield NN works as an associative memory to retrieve one of the embedded patterns from the local data of the input images. The confidence degree means the difference between the retrieved pattern and the input image. When the difference is small the confidence degree is set to be large. The confidence degree works to enhance the CNN operation. By computer simulations, we investigate the basic property of the proposed method and confirm its effectiveness for example of pattern detection.