Rejection sampling is perhaps not a good option, because the assumption does not hold: "the maximum weight isn't too much greater than the average weight" In l

Rejection sampling is perhaps not a good option, because the assumption does not hold: "the maximum weight isn't too much greater than the average weight" In learning algorithms, the distribution of the weights can get piqued.