A new neural network for solving a class of constrained least square problems

A new neural network for solving a class of constrained least square problems is presented. The network is shown to be completely stable and globally convergent to the exact solutions to the constrained least square problem. In contrast to the neural network proposed in the Ref.[1], our new neural network has the following advantages in two major aspects. First, the convergent region of this new network is the whole space R-n. Second, in hardware implementations this new network does not need the expensive analogue multiplier for variables.