My doctorial thesis took a hard look at an emerging field of computational science: the neural network. I was interested in the applications of this technology in my profession: the management of production systems.

After much reading and research, I was convinced that there were numerous applications of this tool. I went on to design a rather different approach to neural networks. Instead of using general purpose neurons and a rather sophisticated learning mechanism, I developed different types of specialized neurons and used a simple network processing algorithm. This approach, by the way, seems to also be going on in biological networks, where specialization of neurons has been discovered.

Instead of letting the learning algorithm do its work, I constructed the network so that it was hard-wired to perform a specific operation, usually optimization. From there, I developed multiple designs that would optimize manufacturing times or minimize routes for material flows.

In order do be able to simulate these I had to develop my own simulation system. Based on C++, I developed a persistence solution and multiple simulation algorithms to be able to test my thoughts.