3

I built a basic neural network in MATLAB. The neural network classifies points on the X-Y axis system into two classes (0 and 1).

(I try to get the function that represents a shape from this photo)

enter image description here

Every so often the values ​​of the points change slightly and some of the points defined in class 1 become class 0, like in this photo.

enter image description here

Is there a way to update the neural network to fit the new data without the time required for retraining?

desertnaut
  • 1,005
  • 10
  • 19
shlomo odem
  • 39
  • 1

1 Answers1

0

If I understand your question, you want it to generalize to be conditioned on an image. If this is correct, you can do this via inserting a separate portion of the model that takes the image as an input, and compresses it into a dense feature vector (can be done many ways- most common is probably cnn) and then merge with the other inputs (in your case x,y pairs) and have the model train on that end-to-end.

mshlis
  • 2,349
  • 7
  • 23
  • If I understood correctly what you mean, you say insert the image as input along with the XY coordinates and the network output will be Is the coordinate inside the black part of the image? And what exactly is "dense feature vector"? Sorry for the ignorance, I'm really new to the field. – shlomo odem Nov 05 '21 at 06:54