Computer Science – Learning
Scientific paper
2011-09-26
Computer Science
Learning
Fixed some typos
Scientific paper
Stochastic gradient descent (SGD) is a simple and popular method to solve stochastic optimization problems which arise in machine learning. For strongly convex problems, its convergence rate was known to be O(log(T)/T), by running SGD for T iterations and returning the average point. However, recent results showed that using a different algorithm, one can get an optimal O(1/T) rate. This might lead one to believe that standard SGD is suboptimal, and maybe should even be replaced as a method of choice. In this paper, we investigate the optimality of SGD in a stochastic setting. We show that for smooth problems, the algorithm attains the optimal O(1/T) rate. However, for non-smooth problems, the convergence rate with averaging might really be \Omega(log(T)/T), and this is not just an artifact of the analysis. On the flip side, we show that a simple modification of the averaging step suffices to recover the O(1/T) rate, and no other change of the algorithm is necessary. We also present experimental results which support our findings, and point out open problems.
No associations
LandOfFree
Making Gradient Descent Optimal for Strongly Convex Stochastic Optimization does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Making Gradient Descent Optimal for Strongly Convex Stochastic Optimization, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Making Gradient Descent Optimal for Strongly Convex Stochastic Optimization will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-689950