A design method for noise canceller using recurrent neural filter <Abstract> This paper presents a design method for noise canceller using recurrent neural filter (RNF). RNF includes the recurrent neural network that has self-feedback at hidden neurons. RNF is divided into several small-scale RNFs connected in parallel (PRNF) to decrease the iterations of calculation for high-speed processing. In addition, the value of learning coefficient is varied dynamically to accelerate the convergence of RNF. Finally, we show the results of numerical simulation and illustrate the effectiveness of the proposed method. |