|
|
Using the AP1000, a distributed-memory machine with 128 processors, it is possible to train a neural net with several mapping schemes and independently of the number of neurons per layer, and the number of layers. Thus, the program is independent of the topology. |
