Fast Training Logistic Regression via Adaptive Sampling

Algorithm 2

AS algorithm.

	Input: Dataset , a weight vector , parameters .
	Output: The vector .
(1)	Initialize: , ,
(2)	Compute , where
	whiledo
(3)
(4)	Sample a random subset with the size of from the set ,
	for eachdo
(5)	Compute
	ifthen
(6)
(7)
	end
	end
	end
(8)	Return