Reconfigurable Recurrent Self-Organizing Maps
Reconfigurable recurrent SOM is unrecognized in the development of neural networks. Separately, software neurons contain fragments of parts of analyzed patterns that have already been successfully identified or used.
The fractal pattern of the previous information is preserved.
These networks are demanding for processing with a GPU cloud.
While RRSOM is in development and testing, I hope it will sometimes be used in the prediction of multidimensional time series.