How to Facilitate Interpretation of Natural Computation Results by Converting Binary Codes of Images Back to Images

This work seeks to illustrate a segment of an in progress research project on the understanding, deciphering and codification of the iconic language of late third millennium Mesopotamian glyptic. The relatively homogeneous corpus of “presentation scenes” involves a number of questions about the logics underlying the structure of each scene and the consequences of changes in the scenes' iconography, with the different relevant meanings. The use of a Self-Organizing Maps-based algorithm in the classification of “presentation scenes” carved on Ur III cylinder seals requires both proper segmenting and codifying systems, and functional methods to interpret the results. This work deals with the problem of obtaining a sound synthetic description of medium-size or large-size SOMs related to a system of relationships between the “presentation scenes” of the examined corpus. The scenes will be reconstructed by an automatic system of interpretation and graphic-textual representation of the SOM codebooks. Such a system appears to be a good means to a quick analysis and a synthetic evaluation of the SOM processing results. Here we will explain the “meta-process” of this Matlab-based program (called SEME) in describing SOM codebooks pertaining to the iconographic classification of the examined Ur III “presentation scenes.”