Closed bengaltiger closed 9 years ago
well, it's not possible to have two input layers in one network using this library, you can either have a big enough input layer to fit all the possible inputs, or maybe training two different network one with each source? If you want you can also create two separate networks, and then connect those two to a third one (networks can project connections to other networks). And then you activate/propagate the networks that you want to use/train... I've never investigated much in using multiple connected networks tho, so I don't know how would that work.
From the sounds of it, it seems like using three networks is exactly what you're looking for @bengaltiger. The inputs to the third would be the outputs of the first two, creating that broken but unified connection that you described. If that would have any benefit above just simply using all the inputs together is another question though.
Is it possible to have multiple input layers? I'm thinking of the case where you have 2 different sources of data as inputs, but a unified target. Initially you pass the inputs through one or more hidden layers (separately) and then to a joint hidden layer. An example would be the "kinship" problem, shown in figure 1 in this paper: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.30.6663