Existing training techniques for spiking neuronal networks tend to be monolithic in nature and scale poorly to larger networks. This paper presents a technique for combining multiple functional neural groupings into a more complex composite network. This is accomplished by ensuring that four axioms hold true for the composite network. The axioms were designed to ensure that incoming signals arrive simultaneously to any component groupings. A number of experiments were conducted in which an algorithm implementing the axioms was used to combine component groupings into more complex networks; these experiments show the practical utility of the technique and reinforce by demonstration the correctness of the axioms.

Artificial intelligence, Constraint programming, Modular neural network, Spiking neural network
Biologically Inspired Cognitive Architectures
School of Computer Science

Bennett, A. (Adam), & White, A. (2018). Synfire circuits: Constraint programming technique for combining functional groupings of spiking neurons. Biologically Inspired Cognitive Architectures. doi:10.1016/j.bica.2018.07.008