according to the paper, the weight value is w = exp(flag * f)
and the f score is the sum of all the weak classification tree score.
so when there is more and more weak classifers, the f is became bigger and bigger
when the value is bigger than 700+, the double value will not be able to hold the result.
any ideas?
according to the paper, the weight value is w = exp(flag * f)
and the f score is the sum of all the weak classification tree score.
so when there is more and more weak classifers, the f is became bigger and bigger
when the value is bigger than 700+, the double value will not be able to hold the result.
any ideas?