Placing spline knots in neural networks using splines as activation functions.
When using feed-forward neural networks with spline activation functions, the quality of approximation depends on the knot placement of spline functions. We demonstrate a method of choosing equidistant knots in each subdivision of the space when an arbitrary initial division is given, in order to ke...
| Published in: | Neurocomputing :an international journal. 17, 3-4 (1997). |
|---|---|
| Main Author: | |
| Format: | Article |
| Language: | English |
| Subjects: |