Control tasks involving dramatic non-linearities, such as decision making, can be challenging for classical design methods. However, autonomous stochastic design methods have proved effective. In particular, Genetic Algorithms (GA) that create phenotypes by the application of genotypes comprising rules are robust and highly scalable. Such encodings are useful for complex applications such as artificial neural net design. This paper outlines an evolutionary algorithm that creates C++ programs which in turn create Artificial Neural Networks (ANNs) that can functionally perform as an exclusive-OR logic gate. Furthermore, the GAs are able to create scalable ANNs robust enough to feature redundancies that allow the network to function despite internal failures.
Modeling of stress-regulated AND (OR) logic gate based on flipping preference of tilted nanomagnet
