Motivation: acquire expert knowledge from examples of expert's problem solving.
Assumption is that it is easier for expert to demonstrate his expertise than to ''tell all he knows``.
Input to the induction algorithm is classified examples (which corresponds to I/O of human expert):
| < f11, f12, ..., f1n > | classification1 |
| < f21, f22, ..., f2m > | classification2 |
Output from the induction algorithm is a decision tree with features labeling interior nodes and classifications labeling leaves.