18 results
Search Results
Now showing 1 - 10 of 18
- ItemConvex optimization with inexact gradients in Hilbert space and applications to elliptic inverse problems(Berlin : Weierstraß-Institut für Angewandte Analysis und Stochastik, 2021) Matyukhin, Vladislav; Kabanikhin, Sergey; Shishlenin, Maxim; Novikov, Nikita; Vasin, Artem; Gasnikov, AlexanderIn this paper we propose the gradient descent type methods to solve convex optimization problems in Hilbert space. We apply it to solve ill-posed Cauchy problem for Poisson equation and make a comparative analysis with Landweber iteration and steepest descent method. The theoretical novelty of the paper consists in the developing of new stopping rule for accelerated gradient methods with inexact gradient (additive noise). Note that up to the moment of stopping the method ``doesn't feel the noise''. But after this moment the noise start to accumulate and the quality of the solution becomes worse for further iterations.
- ItemFlexible modification of Gauss--Newton method and its stochastic extension(Berlin : Weierstraß-Institut für Angewandte Analysis und Stochastik, 2021) Yudin, Nikita; Gasnikov, AlexanderThis work presents a novel version of recently developed Gauss--Newton method for solving systems of nonlinear equations, based on upper bound of solution residual and quadratic regularization ideas. We obtained for such method global convergence bounds and under natural non-degeneracy assumptions we present local quadratic convergence results. We developed stochastic optimization algorithms for presented Gauss--Newton method and justified sub-linear and linear convergence rates for these algorithms using weak growth condition (WGC) and Polyak--Lojasiewicz (PL) inequality. We show that Gauss--Newton method in stochastic setting can effectively find solution under WGC and PL condition matching convergence rate of the deterministic optimization method. The suggested method unifies most practically used Gauss--Newton method modifications and can easily interpolate between them providing flexible and convenient method easily implementable using standard techniques of convex optimization.
- ItemZeroth-order algorithms for smooth saddle-point problems(Berlin : Weierstraß-Institut für Angewandte Analysis und Stochastik, 2021) Sadiev, Abdurakhmon; Beznosikov, Aleksandr; Dvurechensky, Pavel; Gasnikov, AlexanderSaddle-point problems have recently gained an increased attention from the machine learning community, mainly due to applications in training Generative Adversarial Networks using stochastic gradients. At the same time, in some applications only a zeroth-order oracle is available. In this paper, we propose several algorithms to solve stochastic smooth (strongly) convex-concave saddle- point problems using zeroth-order oracles, and estimate their convergence rate and its dependence on the dimension n of the variable. In particular, our analysis shows that in the case when the feasible set is a direct product of two simplices, our convergence rate for the stochastic term is only by a log n factor worse than for the first-order methods. We also consider a mixed setup and develop 1/2th-order methods which use zeroth-order oracle for the minimization part and first-order oracle for the maximization part. Finally, we demonstrate the practical performance of our zeroth-order and 1/2th-order methods on practical problems.
- ItemInexact relative smoothness and strong convexity for optimization and variational inequalities by inexact model(Berlin : Weierstraß-Institut für Angewandte Analysis und Stochastik, 2020) Stonyakin, Fedor; Gasnikov, Alexander; Tyurin, Alexander; Pasechnyuk, Dmitry; Agafonov, Artem; Dvurechensky, Pavel; Dvinskikh, Darina; Artamonov, Sergei; Piskunova, VictoryaIn this paper we propose a general algorithmic framework for first-order methods in optimization in a broad sense, including minimization problems, saddle-point problems and variational inequalities. This framework allows to obtain many known methods as a special case, the list including accelerated gradient method, composite optimization methods, level-set methods, Bregman proximal methods. The idea of the framework is based on constructing an inexact model of the main problem component, i.e. objective function in optimization or operator in variational inequalities. Besides reproducing known results, our framework allows to construct new methods, which we illustrate by constructing a universal conditional gradient method and universal method for variational inequalities with composite structure. These method works for smooth and non-smooth problems with optimal complexity without a priori knowledge of the problem smoothness. As a particular case of our general framework, we introduce relative smoothness for operators and propose an algorithm for VIs with such operator. We also generalize our framework for relatively strongly convex objectives and strongly monotone variational inequalities.
- ItemOn accelerated alternating minimization(Berlin : Weierstraß-Institut für Angewandte Analysis und Stochastik, 2020) Guminov, Sergey; Dvurechensky, Pavel; Gasnikov, AlexanderAlternating minimization (AM) optimization algorithms have been known for a long time and are of importance in machine learning problems, among which we are mostly motivated by approximating optimal transport distances. AM algorithms assume that the decision variable is divided into several blocks and minimization in each block can be done explicitly or cheaply with high accuracy. The ubiquitous Sinkhorn's algorithm can be seen as an alternating minimization algorithm for the dual to the entropy-regularized optimal transport problem. We introduce an accelerated alternating minimization method with a $1/k^2$ convergence rate, where $k$ is the iteration counter. This improves over known bound $1/k$ for general AM methods and for the Sinkhorn's algorithm. Moreover, our algorithm converges faster than gradient-type methods in practice as it is free of the choice of the step-size and is adaptive to the local smoothness of the problem. We show that the proposed method is primal-dual, meaning that if we apply it to a dual problem, we can reconstruct the solution of the primal problem with the same convergence rate. We apply our method to the entropy regularized optimal transport problem and show experimentally, that it outperforms Sinkhorn's algorithm.
- ItemNear-optimal tensor methods for minimizing gradient norm(Berlin : Weierstraß-Institut für Angewandte Analysis und Stochastik, 2020) Dvurechensky, Pavel; Gasnikov, Alexander; Ostroukhov, Petr; Uribe, A. Cesar; Ivanova, AnastasiyaMotivated by convex problems with linear constraints and, in particular, by entropy-regularized optimal transport, we consider the problem of finding approximate stationary points, i.e. points with the norm of the objective gradient less than small error, of convex functions with Lipschitz p-th order derivatives. Lower complexity bounds for this problem were recently proposed in [Grapiglia and Nesterov, arXiv:1907.07053]. However, the methods presented in the same paper do not have optimal complexity bounds. We propose two optimal up to logarithmic factors methods with complexity bounds with respect to the initial objective residual and the distance between the starting point and solution respectively
- ItemOn primal and dual approaches for distributed stochastic convex optimization over networks(Berlin : Weierstraß-Institut für Angewandte Analysis und Stochastik, 2020) Dvinskikh, Darina; Gorbunov, Eduard; Gasnikov, Alexander; Dvurechensky, Alexander; Uribe, César A.We introduce a primal-dual stochastic gradient oracle method for distributed convex optimization problems over networks. We show that the proposed method is optimal in terms of communication steps. Additionally, we propose a new analysis method for the rate of convergence in terms of duality gap and probability of large deviations. This analysis is based on a new technique that allows to bound the distance between the iteration sequence and the optimal point. By the proper choice of batch size, we can guarantee that this distance equals (up to a constant) to the distance between the starting point and the solution.
- ItemAdvances in low-memory subgradient optimization(Berlin : Weierstraß-Institut für Angewandte Analysis und Stochastik, 2020) Dvurechensky, Pavel; Gasnikov, Alexander; Nurminski, Evgeni; Stonyakin, FedorOne of the main goals in the development of non-smooth optimization is to cope with high dimensional problems by decomposition, duality or Lagrangian relaxation which greatly reduces the number of variables at the cost of worsening differentiability of objective or constraints. Small or medium dimensionality of resulting non-smooth problems allows to use bundle-type algorithms to achieve higher rates of convergence and obtain higher accuracy, which of course came at the cost of additional memory requirements, typically of the order of n2, where n is the number of variables of non-smooth problem. However with the rapid development of more and more sophisticated models in industry, economy, finance, et all such memory requirements are becoming too hard to satisfy. It raised the interest in subgradient-based low-memory algorithms and later developments in this area significantly improved over their early variants still preserving O(n) memory requirements. To review these developments this chapter is devoted to the black-box subgradient algorithms with the minimal requirements for the storage of auxiliary results, which are necessary to execute these algorithms. To provide historical perspective this survey starts with the original result of N.Z. Shor which opened this field with the application to the classical transportation problem. The theoretical complexity bounds for smooth and non-smooth convex and quasi-convex optimization problems are briefly exposed in what follows to introduce to the relevant fundamentals of non-smooth optimization. Special attention in this section is given to the adaptive step-size policy which aims to attain lowest complexity bounds. Unfortunately the non-differentiability of objective function in convex optimization essentially slows down the theoretical low bounds for the rate of convergence in subgradient optimization compared to the smooth case but there are different modern techniques that allow to solve non-smooth convex optimization problems faster then dictate lower complexity bounds. In this work the particular attention is given to Nesterov smoothing technique, Nesterov Universal approach, and Legendre (saddle point) representation approach. The new results on Universal Mirror Prox algorithms represent the original parts of the survey. To demonstrate application of non-smooth convex optimization algorithms for solution of huge-scale extremal problems we consider convex optimization problems with non-smooth functional constraints and propose two adaptive Mirror Descent methods. The first method is of primal-dual variety and proved to be optimal in terms of lower oracle bounds for the class of Lipschitz-continuous convex objective and constraints. The advantages of application of this method to sparse Truss Topology Design problem are discussed in certain details. The second method can be applied for solution of convex and quasi-convex optimization problems and is optimal in a sense of complexity bounds. The conclusion part of the survey contains the important references that characterize recent developments of non-smooth convex optimization.
- ItemOptimal decentralized distributed algorithms for stochastic convex optimization(Berlin : Weierstraß-Institut für Angewandte Analysis und Stochastik, 2020) Gorbunov, Eduard; Dvinskikh, Darina; Gasnikov, AlexanderWe consider stochastic convex optimization problems with affine constraints and develop several methods using either primal or dual approach to solve it. In the primal case we use special penalization technique to make the initial problem more convenient for using optimization methods. We propose algorithms to solve it based on Similar Triangles Method with Inexact Proximal Step for the convex smooth and strongly convex smooth objective functions and methods based on Gradient Sliding algorithm to solve the same problems in the non-smooth case. We prove the convergence guarantees in smooth convex case with deterministic first-order oracle. We propose and analyze three novel methods to handle stochastic convex optimization problems with affine constraints: SPDSTM, R-RRMA-AC-SA and SSTM_sc. All methods use stochastic dual oracle. SPDSTM is the stochastic primal-dual modification of STM and it is applied for the dual problem when the primal functional is strongly convex and Lipschitz continuous on some ball. R-RRMA-AC-SA is an accelerated stochastic method based on the restarts of RRMA-AC-SA and SSTM_sc is just stochastic STM for strongly convex problems. Both methods are applied to the dual problem when the primal functional is strongly convex, smooth and Lipschitz continuous on some ball and use stochastic dual first-order oracle. We develop convergence analysis for these methods for the unbiased and biased oracles respectively. Finally, we apply all aforementioned results and approaches to solve decentralized distributed optimization problem and discuss optimality of the obtained results in terms of communication rounds and number of oracle calls per node.
- ItemOracle complexity separation in convex optimization(Berlin : Weierstraß-Institut für Angewandte Analysis und Stochastik, 2020) Ivanova, Anastasiya; Gasnikov, Alexander; Dvurechensky, Pavel; Dvinskikh, Darina; Tyurin, Alexander; Vorontsova, Evgeniya; Pasechnyuk, DmitryUbiquitous in machine learning regularized empirical risk minimization problems are often composed of several blocks which can be treated using different types of oracles, e.g., full gradient, stochastic gradient or coordinate derivative. Optimal oracle complexity is known and achievable separately for the full gradient case, the stochastic gradient case, etc. We propose a generic framework to combine optimal algorithms for different types of oracles in order to achieve separate optimal oracle complexity for each block, i.e. for each block the corresponding oracle is called the optimal number of times for a given accuracy. As a particular example, we demonstrate that for a combination of a full gradient oracle and either a stochastic gradient oracle or a coordinate descent oracle our approach leads to the optimal number of oracle calls separately for the full gradient part and the stochastic/coordinate descent part.