I am new Machine Learning, I would be grateful if someone could help me out. I have read few docs on Neural Nets. I am supposed to design a classification model which is character recognizer. I know that we can train using the cnn and ANN by images. But, I am asked to do an online handwritten character recognition system which has a sequence of (x,y) coordinates which are fetched (time context). I did some preprocessing like normalization, interpolation and smoothing and slant correction and extracted few features like first and second order derivatives at each point and the change in curvature.
How am I supposed to feed them to the Neural Net, Some suggested using RNN. Can anyone give me a sample code snippet to use RNN for taking in the sequence of data but classifying it one class? Should I use some other kind of Neural Net?
Thanks in advance.