Clustering with mean field annealing and unsupervised learning

When neural networks are used to solve a clustering problem, there is often no precise measure. However, in such fields as pattern recognition, a clustering problem is often with an objective function. In this paper, mean field theory neural nets are taken to tackle such a problem. Even when the number of clusters is unknown, an unsupervised neural network with gradient descent can evaluate it. The experimental result is satisfactory.<>