MIT
122
Random Sampling
lGenerate
representative subproblem
lUse
it to estimate solution to whole
»Gives
approximate solution
»May
be quickly repaired to exact solution
lBias
sample toward “important” or “sensitive” parts of problem
lNew
max-flow and min-cut algorithms