Add to Chrome
Add to Firefox
Add to EdgeStochastic Variance Reduced Primal Dual Algorithms for Empirical Composition Optimization
Jul 22, 2019
We consider a generic empirical composition optimization problem, where there are empirical averages present both outside and inside nonlinear loss functions. Such a problem is of interest in various machine learning applications, and cannot be directly solved by standard methods such as stochastic gradient descent (SGD). We take a novel approach to solving this problem by reformulating the original minimization objective into an equivalent min-max objective, which brings out all the empirical averages that are originally inside the nonlinear loss functions. We exploit the rich structures of the reformulated problem and develop a stochastic primal-dual algorithm, SVRPDA-I, to solve the problem efficiently. We carry out extensive theoretical analysis of the proposed algorithm, obtaining the convergence rate, the total computation complexity and the storage complexity. In particular, the algorithm is shown to converge at a linear rate when the problem is strongly convex. Moreover, we also develop an approximate version of the algorithm, SVRPDA-II, which further reduces the memory requirement. Finally, we evaluate the performance of our algorithms on several real-world benchmarks, and experimental results show that the proposed algorithms significantly outperform existing techniques.
