An integrated symbolic and neural network architecture for machine learning in the domain of nuclear engineering
Ephraim Nissan, Hava Siegelmann, Alex Galperin · 2002
On top of FUELCON and NEL, two extant, successful projects in, respectively, expert systems for engineering, and neural networks, we have defined and designed a new phase, meant to greatly increase the significance, for AI, of the combined project with respect to the already recognized merits of the two seed-projects. The NEL symbolic-to-neural conversion schema and language is resorted to in NEURALIZER, a component meant to automatically revise a ruleset, iteration after iteration, within the operation cycle of FUELCON, a generator of families of configurations of fuel assemblies for reloading the core of nuclear reactors.