Multi-Valued "And/Or"-Neural Network

To construct of a multi-layer network in a FPGA, we discuss simplified network constructions. We reexamined neuron functions for back-propagation learning. We made some improvements for the functions, but couldn't achieve drastic reduction. As the results, we abandoned back-propagation learning, and proposed a new neural network, named as AND/OR-neural network, which is derived from the disjunctive normal-form of logical expressions. The network is defined in the binary logic only and has a conclusive learning. The hardware amounts are less than that of the original, and the network can be implemented in a small size FPGA. However, the AND/OR-neural network has not prediction ability. Therefore, to obtain the ability, we expand it to multi-valued type. The extension is done approximately by replacements of logical operators. We discussed the property. We implemented the multi-valued AND/OR-network in 20K gates FPGA, and we could solve 7th-dimensional exclusive-OR problem in order of microsecond. Keywords: Neural network, FPGA, HDL, conclusive learning, disjunctive normal form, multi-dimensiona