A Random Attributed Relational Graph (RARG) or Random Attributed Graph (RAG) is a graph in which vertices and edges are associated with discrete or real-valued random variables. A RARG is an extension of the traditional Random Graph in which only the randomness of the edge connection and vertex presense is modeled. While the instances generated from a Random Graph are conventional graphs for representing strucutures, the instances generated from a Random Attributed Relational Graph are Attributed Relational Graphs for representing attributed structures (see below).

	An ARG is an instance generated from a RARG

Random Attributed Relatinal Graph can be used to model a class of relational data that have common topological or attributive properties. In our current project, we use RARG to model an object model that represents a class of objects, such as face. For more details, please see our part-based object detection web page. For more technical details, please read the following papers:

Dong-Qing Zhang and Shih-Fu Chang, "Learning random attributed relational graph for part-based object detection", Columbia University ADVENT Technical Report #212-2005-6, May, 2005.[pdf]