Title :
Neural network structures with feedback, in binary hypothesis testing
Author :
Halford, Karen W. ; Kazakos, Dimitri ; Pados, Dimitrios ; Papantoni-Kazakos, P.
Author_Institution :
Dept. of Electr. Eng., Virginia Univ., Charlottesville, VA, USA
Abstract :
Two fundamental neural network structures with feedback are considered: a structure containing a fusion center, and a Hopfield network structure. The objective of the two structures is assumed to be binary hypothesis testing. The optimal operations for the Neyman-Pearson criterion are established, for stationary and memoryless hypotheses, and the time evolution of the induced power sequences is developed. The asymptotic performance of the two structures, in terms of asymptotic relative efficiency, is studied and the effects of the feedback are quantified. In addition, the effects of robust data operations per neural element are studied as well. The benefits of feedback and the robust operations are established
Keywords :
feedback; neural nets; statistical analysis; Hopfield network; Neyman-Pearson criterion; asymptotic performance; asymptotic relative efficiency; binary hypothesis testing; feedback; fusion center; memoryless hypotheses; neural network; stationary hypotheses; statistical analysis; Character recognition; Computer networks; Hopfield neural networks; Intelligent networks; Multi-layer neural network; Neural networks; Neurofeedback; Pattern recognition; Robustness; Testing;
Conference_Titel :
Systems, Man, and Cybernetics, 1991. 'Decision Aiding for Complex Systems, Conference Proceedings., 1991 IEEE International Conference on
Conference_Location :
Charlottesville, VA
Print_ISBN :
0-7803-0233-8
DOI :
10.1109/ICSMC.1991.169897
