An unsupervised neural network for machine part recognition with constraint release

In this paper, we provide a study on the learning adaptation of an unsupervised neural network when applied to machine part recognition. The network used is based on an unsupervised learning algorithm called learning by experience (LBE). Here, we modify the network so that whenever it encounters a memory full case, it adopts an approach by releasing the constraint to counteract this effect. Hence, it provides the flexibility for machine part recognition. Simulation results are included.<>