Author(s): Yann LeCun

standard one-hidden layer, fully-connected neural net. this network is replicable. (new net-ff thickin thick0 thick1 ini inj prm)
[CLASS] (packages/gblearn2/net-ff.lsh)

create a new net-ff network.
<thickin>: number of inputs.
<thick0>: size of hidden layer
<thick1>: number of outputs
<ini> <inj>: expected max size of input over which the network
             shbould be replicated.
<prm> an idx1-ddparam in which the parameters will be allocated.