An algorithm for the automatic generation of neural network structures

A backpropagation-based algorithm that dynamically configures the structure of feedforward multilayered neural networks and demonstrates its potential for control applications. The algorithm presents a systematic method for selecting neural network structures according to the complexity of the required mappings. A generate-and-test scheme is employed to evaluate the learning performance of the structure used and modify it accordingly by exploring different alternative structures and selecting the most suitable one. The efficiency of the algorithm is demonstrated using two case studies