| Title |
Novel Artificial Neural Networks Models with Multiple Bias Channels and Its Applications to Signal Process |
| Authors |
하양(Yang Ha) ; 조현철(Hyun Cheol Cho) |
| DOI |
https://doi.org/10.5370/KIEEP.2020.69.1.27 |
| Keywords |
Artificial neural networks; multiple bias; perceptron; recurrent networks; local stability |
| Abstract |
Artificial neural networks technique is significantly applied to solve complicated engineering and scientific problems in the fields of pattern recognition, robot controls, autonomous driving systems, computer games, etc. This paper presents novel neural networks models employed multiple bias input channels particularly for signal processing applications. We design three types of neural networks model including single-layer and multilayer perceptron, and recurrent networks in which multiple bias signals in the first layer are transfered to the next layer. Additionally, an analytical study is carried out to seek local stability, controllability, and observability against the proposed recurrent network model. Lastly, computer simulation is numerically conducted to demonstrate reliability of the proposed 3-layer perceptron model that works to solve mapping problems |