The starting point for the layer concept is the domain-specific language CONNECT. This language had been developed in the 90s, combined procedural, functional and object-oriented paradigms and allowed for readable and complete definitions of connectionist systems. A compiler translated CONNECT specifications into C++ code.
The CONNECT language became part of the NeuroLution system, a system which consisted of hardware and software components. For many standard network types, simulators with comfortable graphical user interfaces were offered. The system could easily be extended by user defined items, because any simulator was characterized by two abstract definitions: a network definition written in the neural network description language CONNECT, and an interface definition based on the GINA library of graphical user interface classes. Also, C++ classes generated from the CONNECT definitions could be exported for the development of custom applications. The NeuroLution PCI board contained up to four SAND neurochips, and accelerated feedforward, RBF, and Kohonen networks.
The concept of the neural network description language CONNECT was later transferred to object-oriented languages and implemented prototypical for C#. The so called CONNECT Framework consisted of a library and a command line tool. From the programmers point of view it extended the C# language by a few attributes and offered a few classes.
The JLayer Framework presented here is a further development of the CONNECT framework and a transfer on the Java programming language. However, it is much more mature than the framework for C#, not least because of the provision of an annotation processor that seamlessly integrates the layer concept into the framework of the Java language.