Recently , potential-based models are gaining considerable interest within the machine learning community . Differing from standard algorithms, these designs define a probability distribution not directly , but by means of a intricate energy mapping . This allows for representing extremely nuanc