This paper provides a review of the article entitled “Scaling Up Stochastic Dual Coordinate Ascent” by K.Tran, S.Hosseini, L.Xiao, T.Finley and M.Bilenko. It reveals positive and negative aspects of the article. Also, in this paper another parallel version of the algorithm, named as UPDA, is introduced. UPDA seems to be efficient and strong algorithm, which can tackle out even problems with several local minimums, what other descent based methods can not do.
AffiliateLabz
Great content! Super high-quality! Keep it up! 🙂