We wish to encourage research into learning algorithms that can handle hard learning problems - hard because there is not enough data, and hard because the target concept is complicated. This year's challenge is based on a classic problem: inducing the language accepted by a randomly generated deterministic finite automaton from a corpus of labeled strings. (See our introduction to DFAs page for definitions.)To this end, we have constructed a table of twelve problems of graded difficulty. Whoever (prior to the end of the competition on 15-Nov-1997) solves a previously unsolved problem harder than any solved problem will win an award of at least $1,024, and will garner eternal fame.