Robustness of automatic target recognition using noniterative neural network
Chia-Lun John Hu · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2005
When the M input training class patterns are represented by N-dimension analog vectors Um, m=1 to M, the output P-bit binary classification vectors Vm, M=1 to M, of a one-layered, feed-forward neural network (OLNN) can be represented geometrically by P dichotomizing hyper-planes going through the origin of the N-dimension Euclidean coordinate in the N-space. In general, all these P planes divide the N-space into 2P hyper-cones. Each cone contains one Um and each cone corresponds to one Vm. Learning of the OLNN is then equivalent to establishing these P planes geometrically in the N-space such that, after the learning, if a test pattern vector T, not necessarily equal to any class pattern Um, falls into the m-th cone (m=1 to 2P) established by these P planes, this T will also be recognized as Vm. The robustness of this recognition is seen now to be equivalent to the geometrically allowed variation range of Um in the m-th cone. This allowable range can be systematically adjusted for each cone during the learning process. This paper reports the optimum method of adjusting these variation ranges such that any unknown T containing environmental noise not included in the training can still be recognized with maximum accuracy.