I'm trying to achieve something relatively simple using an RNN:
1) My input is any value between 0 and 1.
2) My output must be a simple Gaussian Curve, whose parameters depend on the input amplitude. (I am modeling this by using return_sequences=True in Keras and Time Distributed layers)
Currently, my model does not seem to be working well and I fear my approach to the problem is wrong. The reason I have not disclosed my model is because I am looking for completely fresh ideas. I would be very grateful if someone could point me to any references that would help me achieve this task.
I am required to use only a simple RNN model and not an LSTM.
