Building categories from visual experiences

The PercEVAL model is intended to learn concepts through pseudo-visual observations. PercEVAL is a connectionist sub-system of MoHA, an hybrid model designed for knowledge learning in order to understand natural language. The purpose of PercEVAL is to provide an account for categories emergence through perception and understanding of many visual experiences. Categories emergence depends on motivation and language, together with economic constraints of representation. Visual knowledge is represented according to the model of shape diffusion, which grants a stylish way to drop useless details of experiences