A novel algorithm for neural network implementation of Boolean functions and its application to character recognition

Boolean logic is considered to be a good source for classification problems, an area dominated by neural networks. Although quite a few algorithms exist for training and implementing neural networks, no technique exists that can guarantee the transformation of any arbitrary Boolean function to a neural network (28). We have developed an algorithm that accomplishes exactly that. The algorithm is backed up by analytical proof and examples. It is verified using the classic character recognition problem to test its efficacy on Boolean vectors. Thereafter, the base algorithm is extended to a more robust Feature Recognition Algorithm that demonstrates its usefulness for pattern recognition. This algorithm uses piece-wise pattern recognition to provide results in a manner of progressive hierarchy. Results are demonstrated on translated, noisy, scaled, and deformed patterns. Comparisons to existing neural networks are also part of the research. The network's complexity analysis, capacity analysis and entropy analysis is also performed.