Is there a way to train a neural network as $f(x) = {1 \over x}$ precisely?
Asked
Active
Viewed 2,788 times
2 Answers
2
Since the activation function itself can be an inverse function, the answer is yes.
Stephen Rauch
- 1,831
- 11
- 23
- 34
Naman Mehta
- 21
- 1
1
Division by zero is not defined, so at 1/0 its not possible to find the gradient and hence the function cannot be exactly approximated by NN
user80942
- 11
- 1