A Fuzzy-Logic and Multi-Layer Perceptron Hybrid Model Applied to a Real World Steel-Making Problem
Abstract
The most interesting properties of neuro-symbolic hybrid models is that they can allow for the combination of knowledge-based and data-based approaches. We have used such a strategy in a large, complex steel-making problem. We present this problem together with one of the hybrid models that have been tested on it. We chose to fully describe this hybrid because it is representative of such data-knowledge integration properties.