There're three parameters in the Radial Basis Function Networks (RBFN).

  1. Centers of RBFs
  2. Width of RBFs
  3. Weights of RBFs

It's a fact that Weights can be easily updated using a simple Gradient Descent. My question is: Can we optimize Centers and Widths of RBFs using Gradient Descent such that approximation will tend to be better.

Any suggestion is welcome.

  • $\begingroup$ Cannot say for sure but I believe this might lead to rapid overfit since setting widhts and centers of gaussians is usually seen as constraining your model to fully capture variance of your data. $\endgroup$ – Adam Oudad Feb 5 '20 at 16:50

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.