Parallel neural network processing system having an output state determination function

摘要

PROBLEM TO BE SOLVED: To provide the parallel neural network processing system with high versatility having a function capable of highly accurately judging the output state. SOLUTION: A first binary output network means 31 for performing learning by using a first binary teacher signal and transmitting a binary direct output signal for input data and a second binary output network means 35 of performing learning by a converted binary teacher signal for which the first binary teacher signal is code- converted and transmitting a converted binary direct output signal converted by an output conversion means 34 provided with an inverse conversion rule from the converted binary teacher signal to the first binary teacher signal for the input data are parallel connected. This system is constituted of at least the above mentioned means and an output state judgment and selection processing means 39 for selectively transmitting the binary direct output signal or the converted binary direct output signal as a final output signal by matching/non-matching state detection among at least the binary direct output signal, the converted binary direction output signal and an adjacent binary output signal adjacent to the binary direct output signal generated by adjacent output generators 36 and 37 and transmitting a corresponding output stat judgment signal further.