I am a beginner in machine learning and neural networks. I have only used neural networks for classification problems. My aim is to modify it so that it can work for polynomial regression as well. In my problem, I have three inputs and three outputs. My aim is to predict these three outputs based on these three inputs. The outputs are real-valued, and can take positive and negative values.

How should I choose the activation functions? I have only used sigmoid.

  • $\begingroup$ For regression problems we use linear activation. $\endgroup$ – desertnaut May 9 at 17:50
  • $\begingroup$ Try the popular ReLU activation function (or its variant, the leaky ReLU) $\endgroup$ – cogito_ai May 10 at 2:52
  • $\begingroup$ Thank you. I have one more doubt. The three input data are in the range of 100s eg(500,600,700) and the output have mixed units (700, -0.52, 0.45). Should I normalize the input as well as output ? $\endgroup$ – isaac john May 10 at 14:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.