The Neural Network I’ve build with Tensorflow API is consisting of an input [X,Y] and an output [A,B,C,D].
The previous goal of my model is the fit parameters O = [A,B,C,D]. More specifically, the [X,Y] data is created by using a value of O = [A,B,C,D] and [A,B,C,D] need to be predicted after training the model. The code that does this works fine.
This goal has changed into solving the problem of fitting for seperate values of O. So for example, fitting for only A,B,C while not fitting for D while the [X,Y] data is generated with [A,B,C,D]. Ofcourse, this can be done by hardcoding and train a model for 1,2,3 and 4 outputs.
Is it possible to use pretrained neural network, which is trained for fitting [A,B,C,D], for fitting [A,B,C]? This way I don’t have to train the model for fitting each amount of parameters. Hang on please, here is a more visually picture. Hope this helps :
If there is anything more information you need, please let me know. I look forward to hearing from you.