Neural network computational algorithms for least squares estimation problems

Summary form only given, as follows. New computational algorithms employing the Hopfield neural network model are presented for the design of minimum variance estimators. By developing appropriate energy functions and a number representation scheme, systematic procedures for programming the neural network are specified for the two major problems in estimation, namely parameter estimation and state estimation. The programming complexity of the algorithms is discussed and the results of some simulation experiments are presented to demonstrate the performance features.<>