Hand-written digit recognition by one variable stochastic simulated annealing

This paper presents the experimental results of hand-written digit recognition. This neural approach is based on the graph matching, a form of elastic matching, and one-variable stochastic simulated annealing (OVSSA) which is a mean-field annealing algorithm to find a global minimum of a real valued function. Experimental results show that this method is insensitive to displacement, scale changes and rotation, and promising to hand-written character recognition.