Who I want to a neural network model of end - to - end
(1) model data (hypothesis is A side) at the end of the if is x
(2) on the other side of the model (assuming that is B side) data is y
Through the study of neural network model to the corresponding feature vector dimension N: vec (x) and vec (y)

Now I'd like to have A new feature with vec T (x) for joining together, as A final vector said
T is a through the study of other methods have good data vectors,
X is the data now, T is the eigenvector of learning the other methods
Y is data
I want to ask how T deal with x fusion, give a code based on keras,