Adaptive basis functions. Alternatively, we might make a function y(x) from basis functions that depend on additional parameters included in the vector w. In a two-layer feed forward neural network with nonlinear hidden units and a linear output, the function can be written
![463_921f266d-770a-47c8-8782-878269dc1ae2.png](https://secure.tutorsglobe.com/CMSImages/463_921f266d-770a-47c8-8782-878269dc1ae2.png)