This paper is a continuation of early work into the Fundamental Universal Learning Patterns (FULPs) Framework, a biologically-aligned diverse intelligence inspired artificial intelligence training paradigm which synthesizes how all living organisms learn and develop in sequential stages, known as a FULP. Up to this point, the eight stage framework has had a working prototype established for the first five FULPs, with the sixth being built during early adaptive cellular automata application experiments, with the seventh currently in development under the same area of study. Here, the eighth is fully developed and engineered to see how an initial test of the final stage of the framework would react and what results it would produce. This stage was inspired by Graph Neural Network (GNN) and cellular automata neighborhood architecture as well as other organic analogs. Tests were performed on basic first architectures, mathematically sophisticated methods, a combined final system, synthetic data, as well as real world data from North-West University. The initial results show a bimodal relationship within the current architecture, proving that in certain cases it can outperform a traditional GNN, but in others it falls very far behind.
William V. Fullerton (Sun,) studied this question.