Increasingly, potential-based frameworks are securing considerable interest within the computational intelligence community . Differing from traditional deep learning architectures , these designs specify a probability set not directly , but by means of a complex potential mapping . This allows