Recurrent neural networks and filters adaptation with stability control

Recurrent Neural Networks and linear Recursive Filters can be adapted on-line but sometimes with instability problems. Stability control techniques exist for the linear case but they are either computationally expensive or non-robust. For the nonlinear case, stability control is simply never done. This paper presents a new stability control method for ITR adaptive filters that makes possible to continually adapt the coefficients with no need of stability test or poles projection. This method can be applied to various realizations: direct forms, cascade or parallel of second order sections, lattice form. It can be implemented to adapt a simple ER adaptive filter or a locally recurrent neural network such as the LR-MLP with improved performance over other techniques and over not controlling the stability.