MIT
103
Rare Events
l
When
FAIL(
p
)
too small, takes too long
to collect sufficient statistics
l
Solution: skew trials to make interesting
event more likely
l
But in a way that let’s you recover
original probability
