Alternative simulation annealing processes for global optimization in neural networks
T.J. Guillerm, Neil E. Cotter · 1991
Summary form only given, as follows. The simulated annealing method is a tool for finding the global minima of a performance measure function. It is accomplished by constraining the probability distribution of the process to be a Gibbs distribution associated with the measure to be minimized. The only parameter upon which the convergence depends is the cooling schedule of the Gibbs temperature. Alternative distributions have been derived along with the cooling schedules for convergence to a global minimum. A global measure of performance was defined. It was concluded that an algorithm using simple multiplicative and additive functions will perform faster on a computer than an algorithm using more complicated functions.>