arXiv subjects

Tianyi Lin

Publications and source records attributed to Tianyi Lin.

68 records · Page 4Linked to original sources

arXiv2020

An ADMM-Based Interior-Point Method for Large-Scale Linear Programming

We propose a new framework to implement interior point method (IPM) to solve very large linear programs (LP). Traditional IPMs typically use Newton's method to approximately solve a subproblem that aims to minimize a log-barrier penalty function at each iteration. Due its connection to Newton's method, IPM is often classified as second-order method -- a genre that is attached with stability and accuracy at the expense of scalability. Indeed, computing a Newton step amounts to solving a large linear system, which can be efficiently implemented if the input data are reasonably-sized and/or sparse and/or well-structured. However, in case the above premises fail, then the challenge still stands on the way for a traditional IPM. To deal with this challenge, one approach is to apply the iterative procedure, such as preconditioned conjugate gradient method, to solve the linear system. Since the linear system is different each iteration, it is difficult to find good pre-conditioner to achieve the overall solution efficiency. In this paper, an alternative approach is proposed. Instead of applying Newton's method, we resort to the alternating direction method of multipliers (ADMM) to approximately minimize the log-barrier penalty function at each iteration, under the framework of primal-dual path-following for a homogeneous self-dual embedded LP model. The resulting algorithm is an ADMM-Based Interior Point Method, abbreviated as ABIP in this paper. The new method inherits stability from IPM, and scalability from ADMM. Because of its self-dual embedding structure, ABIP is set to solve any LP without requiring prior knowledge about its feasibility. We conduct extensive numerical experiments testing ABIP with large-scale LPs from NETLIB and machine learning applications. The results demonstrate that ABIP compares favorably with existing LP solvers including SDPT3, MOSEK, DSDP-CG and SCS.

math.OC↗

arXiv2020

A Unified Adaptive Tensor Approximation Scheme to Accelerate Composite Convex Optimization

In this paper, we propose a unified two-phase scheme to accelerate any high-order regularized tensor approximation approach on the smooth part of a composite convex optimization model. The proposed scheme has the advantage of not needing to assume any prior knowledge of the Lipschitz constants for the gradient, the Hessian and/or high-order derivatives. This is achieved by tuning the parameters used in the algorithm \textit{adaptively} in its process of progression, which has been successfully incorporated in high-order nonconvex optimization (CartisGouldToint2018, Birgin-Gardenghi-Martinez-Santos-Toint-2017). In general, we show that the adaptive high-order method has an iteration bound of $O\left( 1 / ε^{1/(p+1)} \right)$ if the first $p$-th order derivative information is used in the approximation, which has the same iteration complexity as in that of the nonadaptive version in (Baes-2009, Nesterov-2018) where the Lipschitz constants are assumed to be known and the subproblems are assumed to be solved exactly. Thus, our results partially address the problem of incorporating adaptive strategies into the high-order {\it accelerated} methods raised by Nesterov in (Nesterov-2018), although our strategies cannot assure the convexity of the auxiliary problem and such adaptive strategies are already popular in high-order nonconvex optimization (CartisGouldToint2018, Birgin-Gardenghi-Martinez-Santos-Toint-2017). Our numerical experiment results show a clear effect of real acceleration displayed in the adaptive Newton's method with cubic regularization on a set of regularized logistic regression instances.

math.OC↗

arXiv2019

Improved Oracle Complexity of Variance Reduced Methods for Nonsmooth Convex Stochastic Composition Optimization

We consider the nonsmooth convex composition optimization problem where the objective is a composition of two finite-sum functions and analyze stochastic compositional variance reduced gradient (SCVRG) methods for them. SCVRG and its variants have recently drawn much attention given their edge over stochastic compositional gradient descent (SCGD); but the theoretical analysis exclusively assumes strong convexity of the objective, which excludes several important examples such as Lasso, logistic regression, principle component analysis and deep neural nets. In contrast, we prove non-asymptotic incremental first-order oracle (IFO) complexity of SCVRG or its novel variants for nonsmooth convex composition optimization and show that they are provably faster than SCGD and gradient descent. More specifically, our method achieves the total IFO complexity of $O\left((m+n)\log\left(1/ε\right)+1/ε^3\right)$ which improves that of $O\left(1/ε^{3.5}\right)$ and $O\left((m+n)/\sqrtε\right)$ obtained by SCGD and accelerated gradient descent (AGD) respectively. Experimental results confirm that our methods outperform several existing methods, e.g., SCGD and AGD, on sparse mean-variance optimization problem.

math.OC↗

arXiv2018

Sparsemax and Relaxed Wasserstein for Topic Sparsity

Topic sparsity refers to the observation that individual documents usually focus on several salient topics instead of covering a wide variety of topics, and a real topic adopts a narrow range of terms instead of a wide coverage of the vocabulary. Understanding this topic sparsity is especially important for analyzing user-generated web content and social media, which are featured in the form of extremely short posts and discussions. As topic sparsity of individual documents in online social media increases, so does the difficulty of analyzing the online text sources using traditional methods. In this paper, we propose two novel neural models by providing sparse posterior distributions over topics based on the Gaussian sparsemax construction, enabling efficient training by stochastic backpropagation. We construct an inference network conditioned on the input data and infer the variational distribution with the relaxed Wasserstein (RW) divergence. Unlike existing works based on Gaussian softmax construction and Kullback-Leibler (KL) divergence, our approaches can identify latent topic sparsity with training stability, predictive performance, and topic coherence. Experiments on different genres of large text corpora have demonstrated the effectiveness of our models as they outperform both probabilistic and neural methods.

cs.LG↗

arXiv2018

On the Iteration Complexity Analysis of Stochastic Primal-Dual Hybrid Gradient Approach with High Probability

In this paper, we propose a stochastic Primal-Dual Hybrid Gradient (PDHG) approach for solving a wide spectrum of regularized stochastic minimization problems, where the regularization term is composite with a linear function. It has been recognized that solving this kind of problem is challenging since the closed-form solution of the proximal mapping associated with the regularization term is not available due to the imposed linear composition, and the per-iteration cost of computing the full gradient of the expected objective function is extremely high when the number of input data samples is considerably large. Our new approach overcomes these issues by exploring the special structure of the regularization term and sampling a few data points at each iteration. Rather than analyzing the convergence in expectation, we provide the detailed iteration complexity analysis for the cases of both uniformly and non-uniformly averaged iterates with high probability. This strongly supports the good practical performance of the proposed approach. Numerical experiments demonstrate that the efficiency of stochastic PDHG, which outperforms other competing algorithms, as expected by the high-probability convergence analysis.

cs.LG↗

arXiv2018

Stochastic Primal-Dual Proximal ExtraGradient Descent for Compositely Regularized Optimization

We consider a wide range of regularized stochastic minimization problems with two regularization terms, one of which is composed with a linear function. This optimization model abstracts a number of important applications in artificial intelligence and machine learning, such as fused Lasso, fused logistic regression, and a class of graph-guided regularized minimization. The computational challenges of this model are in two folds. On one hand, the closed-form solution of the proximal mapping associated with the composed regularization term or the expected objective function is not available. On the other hand, the calculation of the full gradient of the expectation in the objective is very expensive when the number of input data samples is considerably large. To address these issues, we propose a stochastic variant of extra-gradient type methods, namely \textsf{Stochastic Primal-Dual Proximal ExtraGradient descent (SPDPEG)}, and analyze its convergence property for both convex and strongly convex objectives. For general convex objectives, the uniformly average iterates generated by \textsf{SPDPEG} converge in expectation with $O(1/\sqrt{t})$ rate. While for strongly convex objectives, the uniformly and non-uniformly average iterates generated by \textsf{SPDPEG} converge with $O(\log(t)/t)$ and $O(1/t)$ rates, respectively. The order of the rate of the proposed algorithm is known to match the best convergence rate for first-order stochastic algorithms. Experiments on fused logistic regression and graph-guided regularized logistic regression problems show that the proposed algorithm performs very efficiently and consistently outperforms other competing algorithms.

cs.LG↗

arXiv2018

Global Convergence of Unmodified 3-Block ADMM for a Class of Convex Minimization Problems

The alternating direction method of multipliers (ADMM) has been successfully applied to solve structured convex optimization problems due to its superior practical performance. The convergence properties of the 2-block ADMM have been studied extensively in the literature. Specifically, it has been proven that the 2-block ADMM globally converges for any penalty parameter $γ>0$. In this sense, the 2-block ADMM allows the parameter to be free, i.e., there is no need to restrict the value for the parameter when implementing this algorithm in order to ensure convergence. However, for the 3-block ADMM, Chen \etal \cite{Chen-admm-failure-2013} recently constructed a counter-example showing that it can diverge if no further condition is imposed. The existing results on studying further sufficient conditions on guaranteeing the convergence of the 3-block ADMM usually require $γ$ to be smaller than a certain bound, which is usually either difficult to compute or too small to make it a practical algorithm. In this paper, we show that the 3-block ADMM still globally converges with any penalty parameter $γ>0$ if the third function $f_3$ in the objective is smooth and strongly convex, and its condition number is in $[1,1.0798)$, besides some other mild conditions. This requirement covers an important class of problems to be called regularized least squares decomposition (RLSD) in this paper.

math.OC↗

arXiv2018

Structured Nonconvex and Nonsmooth Optimization: Algorithms and Iteration Complexity Analysis

Nonconvex and nonsmooth optimization problems are frequently encountered in much of statistics, business, science and engineering, but they are not yet widely recognized as a technology in the sense of scalability. A reason for this relatively low degree of popularity is the lack of a well developed system of theory and algorithms to support the applications, as is the case for its convex counterpart. This paper aims to take one step in the direction of disciplined nonconvex and nonsmooth optimization. In particular, we consider in this paper some constrained nonconvex optimization models in block decision variables, with or without coupled affine constraints. In the case of without coupled constraints, we show a sublinear rate of convergence to an $ε$-stationary solution in the form of variational inequality for a generalized conditional gradient method, where the convergence rate is shown to be dependent on the Hölderian continuity of the gradient of the smooth part of the objective. For the model with coupled affine constraints, we introduce corresponding $ε$-stationarity conditions, and apply two proximal-type variants of the ADMM to solve such a model, assuming the proximal ADMM updates can be implemented for all the block variables except for the last block, for which either a gradient step or a majorization-minimization step is implemented. We show an iteration complexity bound of $O(1/ε^2)$ to reach an $ε$-stationary solution for both algorithms. Moreover, we show that the same iteration complexity of a proximal BCD method follows immediately. Numerical results are provided to illustrate the efficacy of the proposed algorithms for tensor robust PCA.

math.OC↗

arXiv2017

A Unified Scheme to Accelerate Adaptive Cubic Regularization and Gradient Methods for Convex Optimization

In this paper we propose a unified two-phase scheme for convex optimization to accelerate: (1) the adaptive cubic regularization methods with exact/inexact Hessian matrices, and (2) the adaptive gradient method, without any knowledge of the Lipschitz constants for the gradient or the Hessian. This is achieved by tuning the parameters used in the algorithm $\textit{adaptively}$ in its process of progression, which can be viewed as a relaxation over the existing algorithms in the literature. Under the assumption that the sub-problems can be solved approximately, we establish overall iteration complexity bounds for three newly proposed algorithms to obtain an $ε$-approximate solution. Specifically, we show that the adaptive cubic regularization methods with the exact/inexact Hessian matrix both achieve an iteration complexity in the order of $O\left( 1 / ε^{1/3} \right)$, which matches that of the original accelerated cubic regularization method presented in [Nesterov-2008-Accelerating] assuming the availability of the exact Hessian information and the Lipschitz constants, and that the sub-problems are solved to optimality. Under the same two-phase adaptive acceleration framework, the gradient method achieves an iteration complexity in the order $O\left( 1 / ε^{1/2} \right)$, which is known to be best possible (cf. Nesterov-2013-Introductory). Our numerical experiment results show a clear effect of acceleration displayed in the adaptive Newton method with cubic regularization on a set of regularized logistic regression instances.

math.OC↗

arXiv2017

Distributed Linearized Alternating Direction Method of Multipliers for Composite Convex Consensus Optimization

Given an undirected graph $\mathcal{G}=(\mathcal{N},\mathcal{E})$ of agents $\mathcal{N}=\{1,\ldots,N\}$ connected with edges in $\mathcal{E}$, we study how to compute an optimal decision on which there is consensus among agents and that minimizes the sum of agent-specific private convex composite functions $\{Φ_i\}_{i\in\mathcal{N}}$ while respecting privacy requirements, where $Φ_i\triangleq ξ_i + f_i$ belongs to agent-$i$. Assuming only agents connected by an edge can communicate, we propose a distributed proximal gradient method DPGA for consensus optimization over both unweighted and weighted static (undirected) communication networks. In one iteration, each agent-$i$ computes the prox map of $ξ_i$ and gradient of $f_i$, and this is followed by local communication with neighboring agents. We also study its stochastic gradient variant, SDPGA, which can only access to noisy estimates of $\nabla f_i$ at each agent-$i$. This computational model abstracts a number of applications in distributed sensing, machine learning and statistical inference. We show ergodic convergence in both sub-optimality error and consensus violation for DPGA and SDPGA with rates $\mathcal{O}(1/t)$ and $\mathcal{O}(1/\sqrt{t})$, respectively.

math.OC↗

arXiv2015

An Extragradient-Based Alternating Direction Method for Convex Minimization

In this paper, we consider the problem of minimizing the sum of two convex functions subject to linear linking constraints. The classical alternating direction type methods usually assume that the two convex functions have relatively easy proximal mappings. However, many problems arising from statistics, image processing and other fields have the structure that while one of the two functions has easy proximal mapping, the other function is smoothly convex but does not have an easy proximal mapping. Therefore, the classical alternating direction methods cannot be applied. To deal with the difficulty, we propose in this paper an alternating direction method based on extragradients. Under the assumption that the smooth function has a Lipschitz continuous gradient, we prove that the proposed method returns an $ε$-optimal solution within $O(1/ε)$ iterations. We apply the proposed method to solve a new statistical model called fused logistic regression. Our numerical experiments show that the proposed method performs very well when solving the test problems. We also test the performance of the proposed method through solving the lasso problem arising from statistics and compare the result with several existing efficient solvers for this problem; the results are very encouraging indeed.

math.OC↗

arXiv2015

On the Sublinear Convergence Rate of Multi-Block ADMM

The alternating direction method of multipliers (ADMM) is widely used in solving structured convex optimization problems. Despite of its success in practice, the convergence properties of the standard ADMM for minimizing the sum of $N$ $(N\geq 3)$ convex functions with $N$ block variables linked by linear constraints, have remained unclear for a very long time. In this paper, we present convergence and convergence rate results for the standard ADMM applied to solve $N$-block $(N\geq 3)$ convex minimization problem, under the condition that one of these functions is convex (not necessarily strongly convex) and the other $N-1$ functions are strongly convex. Specifically, in that case the ADMM is proven to converge with rate $O(1/t)$ in a certain ergodic sense, and $o(1/t)$ in non-ergodic sense, where $t$ denotes the number of iterations. As a by-product, we also provide a simple proof for the $O(1/t)$ convergence rate of two-block ADMM in terms of both objective error and constraint violation, without assuming any condition on the penalty parameter and strong convexity on the functions.

math.OC↗

arXiv2015

On the Global Linear Convergence of the ADMM with Multi-Block Variables

The alternating direction method of multipliers (ADMM) has been widely used for solving structured convex optimization problems. In particular, the ADMM can solve convex programs that minimize the sum of $N$ convex functions with $N$-block variables linked by some linear constraints. While the convergence of the ADMM for $N=2$ was well established in the literature, it remained an open problem for a long time whether or not the ADMM for $N \ge 3$ is still convergent. Recently, it was shown in [3] that without further conditions the ADMM for $N\ge 3$ may actually fail to converge. In this paper, we show that under some easily verifiable and reasonable conditions the global linear convergence of the ADMM when $N\geq 3$ can still be assured, which is important since the ADMM is a popular method for solving large scale multi-block optimization models and is known to perform very well in practice even when $N\ge 3$. Our study aims to offer an explanation for this phenomenon.

math.OC↗

arXiv2015

Iteration Complexity Analysis of Multi-Block ADMM for a Family of Convex Minimization without Strong Convexity

The alternating direction method of multipliers (ADMM) is widely used in solving structured convex optimization problems due to its superior practical performance. On the theoretical side however, a counterexample was shown in [7] indicating that the multi-block ADMM for minimizing the sum of $N$ $(N\geq 3)$ convex functions with $N$ block variables linked by linear constraints may diverge. It is therefore of great interest to investigate further sufficient conditions on the input side which can guarantee convergence for the multi-block ADMM. The existing results typically require the strong convexity on parts of the objective. In this paper, we present convergence and convergence rate results for the multi-block ADMM applied to solve certain $N$-block $(N\geq 3)$ convex minimization problems without requiring strong convexity. Specifically, we prove the following two results: (1) the multi-block ADMM returns an $ε$-optimal solution within $O(1/ε^2)$ iterations by solving an associated perturbation to the original problem; (2) the multi-block ADMM returns an $ε$-optimal solution within $O(1/ε)$ iterations when it is applied to solve a certain sharing problem, under the condition that the augmented Lagrangian function satisfies the Kurdyka-Lojasiewicz property, which essentially covers most convex optimization models except for some pathological cases.

math.OC↗