A universal optimisation network

An optimization algorithm was developed that causes the simultaneous convergence of a large number of parameters determining the value of a scaler cost function. The procedure is iterative and stochastic, and tends to avoid local extrema. It is shown that the number of iterations required for convergence of the cost function increases linearly with the number of parameters. The procedure is universal in that it can be applied without modification to a large variety of optimization problems. Several examples are discussed, and results of computer simulations are presented.<
>