Skip to Main Content
This paper describes a scalable XML-based approach to store data and configurations of artificial neural networks (ANNs) for the purpose of documentation and exchange between scientific and production tools. After a brief summary on ANNs and how their information is currently stored, some use cases for data related to an ANN are discussed. Derived from these use cases, a specification for a flexible data storage solution is described. A subsequent section presents a complementary approach to store configuration information that also may be scaled to specific needs. Both specifications build up the XML language ANNSL (artificial neural network specification language).