.png){kind=link}
Python (sklearn-based) implementation that explores how different heuristics and parameters impact an RBFNN.
A brief analysis of the results is provided in Portuguese. It was submitted as an assignment of a graduate course named Connectionist Artificial Intelligence at UFSC, Brazil.
In short, three heuristics (out of my head) are evaluated to find out what is the (least worst) way to calculate sigma, an important parameter for RBF. Then, the best-performing heuristic is used to test how many neurons are needed in the hidden layer to get close to 1 f1-score at classifying data from a 2D dataset.
| True Data | RBF Prediction using 25 neurons |
|---|---|
 |
 |