M. Kunze et J. Steffens, THE NEURAL-NETWORK OBJECTS, Nuclear instruments & methods in physics research. Section A, Accelerators, spectrometers, detectors and associated equipment, 389(1-2), 1997, pp. 12-15
The neural network objects (NNO) is a C++ class library that contains
the most popular standard neural networks together with self-organizin
g incremental models. The implementation takes full advantage of the O
O-paradigm, i.e. all models are derived from the same abstract base cl
asses which deal with the data structures and learning algorithms at o
ne central place. All NNO are persistent. After training they save the
mselves to permanent store in a computer independent format and may be
re-activated for further training or recall cycles, even on a differe
nt computer.