Neural networks for optimization of nonquadratic cost functions with application to adaptive signal processing

A new neural adaptive filtering structure was proposed by the authors (1990-1991), based on a least-squares (LS) performance function of errors. This structure is generalized and a neural adaptive finite impulse response (FIR) filter is designed whose performance function is expressed in the general non-LS form. The proposed neural filter is shown to compute in real time the optimal set of the programmable weights for general non-LS cost functions. As a consequence it features excellent tracking capabilities and is effective for online applications where fast adaptation speed is required. It is also shown that for some common non-LS cost functions, the neural structures proposed can be implemented on relatively simple electronic circuits that can be fully integrated in MOS VLSI technology