Instance Based Learning

- Before we go on, we need some definitions:
- Regression means approximating a real-valued target function.
- Residual is the error $\hat{f}(x) - f(x)$ in approximating the target function.
- Kernel function is the function of distance that is used to determine the weight of each training example. That is, $K$ where \[ w_i = K(d(x_i, x_q))\]

8 of 18