Fig. 2From: From big data to smart data: a sample gradient descent approach for machine learningCost_function_before_samplingBack to article page