Application of perceptron to the detecting of particle motion

This paper consists of two parts: an application of a multilayer perceptron model to the detecting of particle motion and a method of expediting the study of the multilayer perceptron training procedure. In the first part, the motion of a single particle or multiple particles is detected by identifying the reflection symmetry of two concatenated image frames. The accuracy will be higher than 90, even though a certain amount of noise exists. In the second part, a convergent index parameter formularized to evaluate the distance of the network inner state from the Boltzmann distribution is put forward for the measurement of the training procedure convergency. With the convergent index, it becomes feasible to study the influence of the network parameters (sizes of different layers, target assigning, etc.) on the training procedure before the network has reached a stable state; thus time is saved. This method is used in the designing of the network.© (1991) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.