Existing convergence analyses of Q-learning mostly focus on the vanilla stochastic gradient descent (SGD) type of updates. Despite the Adaptive Moment Estimation (Adam) has been commonly used for practical Q-learning algorithms, there has not been any convergence guarantee provided for Q-learning with such type of updates. In this paper, we first characterize the convergence rate for Q-AMSGrad, which is the Q-learning algorithm with AMSGrad update (a commonly adopted alternative of Adam for theoretical analysis). To further improve the performance, we propose to incorporate the momentum restart scheme to Q-AMSGrad, resulting in the so-called Q-AMSGradR algorithm. The convergence rate of Q-AMSGradR is also established. Our experiments on a linear quadratic regulator problem show that the two proposed Q-learning algorithms outperform the vanilla Q-learning with SGD updates. The two algorithms also exhibit significantly better performance than the DQN learning method over a batch of Atari 2600 games.
Restart techniques are common in gradient-free optimization to deal with multimodal functions. Partial warm restarts are also gaining popularity in gradient-based optimization to improve the rate of convergence in accelerated gradient schemes to deal with ill-conditioned functions. In this paper, we propose a simple warm restart technique for stochastic gradient descent to improve its anytime performance when training deep neural networks. We empirically study its performance on the CIFAR-10 and CIFAR-100 datasets, where we demonstrate new state-of-the-art results at 3.14% and 16.21%, respecti…
University of Utah（犹他大学）；Rice University（莱斯大学）；National University of Defense Technology（国防科技大学）；University of California, Los Angeles（加州大学洛杉矶分校）
Stochastic gradient descent (SGD) with constant momentum and its variants such as Adam are the optimization algorithms of choice for training deep neural networks (DNNs). Since DNN training is incredibly computationally expensive, there is great interest in speeding up the convergence. Nesterov accelerated gradient (NAG) improves the convergence rate of gradient descent (GD) for convex optimization using a specially designed momentum; however, it accumulates error when an inexact gradient is used (such as in SGD), slowing convergence at best and diverging at worst. In this paper, we propose Sc…
Monash University（莫纳什大学）；University of Queensland（昆士兰大学）；University of Waterloo（滑铁卢大学）
We present Nesterov-type acceleration techniques for Alternating Least Squares (ALS) methods applied to canonical tensor decomposition. While Nesterov acceleration turns gradient descent into an optimal first-order method for convex problems by adding a momentum term with a specific weight sequence, a direct application of this method and weight sequence to ALS results in erratic convergence behaviour. This is so because the tensor decomposition problem is non-convex and ALS is accelerated instead of gradient descent. Instead, we consider various restart mechanisms and suitable choices of mome…
When to restart? Exploring escalating restarts on convergence
Ericsson Research, Ericsson AB, Sweden（爱立信研究, 爱立信公司, 瑞典）；KTH Royal Institute of Technology, Stockholm, Sweden（皇家理工学院）；RISE Research Institutes of Sweden, Stockholm, Sweden（瑞典RISE研究机构）
Provable and Practical Online Learning Rate Adaptation with Hypergradient Descent
Stanford University（斯坦福大学）；ICME, Stanford University（斯坦福大学ICME）；Department of Mathematics, Stanford University（斯坦福大学数学系）；Department of Management Science and Engineering, Stanford University（斯坦福大学管理科学与工程学院）
We propose new restarting strategies for the accelerated coordinate descent method. Our main contribution is to show that for a well chosen sequence of restarting times, the restarted method has a nearly geometric rate of convergence. A major feature of the method is that it can take profit of the local quadratic error bound of the objective function without knowing the actual value of the error bound. We also show that under the more restrictive assumption that the objective function is strongly convex, any fixed restart period leads to a geometric rate of convergence. Finally, we illustrate …
A theoretical, and potentially also practical, problem with stochastic gradient descent is that trajectories may escape to infinity. In this note, we investigate uniform boundedness properties of iterates and function values along the trajectories of the stochastic gradient descent algorithm and its important momentum variant. Under smoothness and $R$-dissipativity of the loss function, we show that broad families of step-sizes, including the widely used step-decay and cosine with (or without) restart step-sizes, result in uniformly bounded iterates and function values. Several important appli…
INRIA Lille（法国国家信息与自动化研究所里尔中心）；Politecnico di Milano（米兰理工大学）
In this paper, we propose a novel approach to automatically determine the batch size in stochastic gradient descent methods. The choice of the batch size induces a trade-off between the accuracy of the gradient estimate and the cost in terms of samples of each update. We propose to determine the batch size by optimizing the ratio between a lower bound to a linear or quadratic Taylor approximation of the expected improvement and the number of samples used to estimate the gradient. The performance of the proposed approach is empirically compared with related methods on popular classification tas…
The deep learning algorithm has achieved great success in the field of computer vision, but some studies have pointed out that the deep learning model is vulnerable to attacks adversarial examples and makes false decisions. This challenges the further development of deep learning, and urges researchers to pay more attention to the relationship between adversarial examples attacks and deep learning security. This work focuses on adversarial examples, optimizes the generation of adversarial examples from the view of adversarial robustness, takes the perturbations added in adversarial examples as…
University of Oxford（牛津大学）；University of British Columbia（不列颠哥伦比亚大学）
We introduce a general method for improving the convergence rate of gradient-based optimizers that is easy to implement and works well in practice. We demonstrate the effectiveness of the method in a range of optimization problems by applying it to stochastic gradient descent, stochastic gradient descent with Nesterov momentum, and Adam, showing that it significantly reduces the need for the manual tuning of the initial learning rate for these commonly used algorithms. Our method works by dynamically updating the learning rate during optimization using the gradient with respect to the learning…
An Approach to Stable Gradient Descent Adaptation of Higher-Order Neural Units
CTU in Prague（捷克布拉格技术大学）；Tohoku University Graduate School of Medicine（东北大学医学研究生院）；Graduate School of Biomedical Engineering, Tohoku University（东北大学生物医学工程研究生院）
University of Maryland, College Park（马里兰大学帕克分校）；University of Siegen（锡根大学）
State-of-the-art adversarial attacks on neural networks use expensive iterative methods and numerous random restarts from different initial points. Iterative FGSM-based methods without restarts trade off performance for computational efficiency because they do not adequately explore the image space and are highly sensitive to the choice of step size. We propose a variant of Projected Gradient Descent (PGD) that uses a random step size to improve performance without resorting to expensive random restarts. Our method, Wide Iterative Stochastic crafting (WITCHcraft), achieves results superior to …
Variance reduction techniques are popular in accelerating gradient descent and stochastic gradient descent for optimization problems defined on both Euclidean space and Riemannian manifold. In this paper, we further improve on existing variance reduction methods for non-convex Riemannian optimization, including R-SVRG and R-SRG/R-SPIDER with batch size adaptation. We show that this strategy can achieve lower total complexities for optimizing both general non-convex and gradient dominated functions under both finite-sum and online settings. As a result, we also provide simpler convergence analy…
Decoupling Variance and Scale-Invariant Updates in Adaptive Gradient Descent for Unified Vector and Matrix Optimization
Department of Computer Science, Purdue University, West Lafayette, USA（计算机科学系，普渡大学，西拉法叶，美国）；Edwardson School of Industrial Engineering, Purdue University, West Lafayette, USA（工业工程学院，普渡大学，西拉法叶，美国）
Carnegie Mellon University（卡内基梅隆大学）；University of Southern California（南加州大学）；Facebook Artificial Intelligence Research（Facebook人工智能研究院）
We consider the problem of learning a one-hidden-layer neural network with non-overlapping convolutional layer and ReLU activation, i.e., $f(\mathbf{Z}, \mathbf{w}, \mathbf{a}) = \sum_j a_jσ(\mathbf{w}^T\mathbf{Z}_j)$, in which both the convolutional weights $\mathbf{w}$ and the output weights $\mathbf{a}$ are parameters to be learned. When the labels are the outputs from a teacher network of the same architecture with fixed weights $(\mathbf{w}^*, \mathbf{a}^*)$, we prove that with Gaussian input $\mathbf{Z}$, there is a spurious local minimizer. Surprisingly, in the presence of the spurious …
Stochastic convex optimization is a basic and well studied primitive in machine learning. It is well known that convex and Lipschitz functions can be minimized efficiently using Stochastic Gradient Descent (SGD). The Normalized Gradient Descent (NGD) algorithm, is an adaptation of Gradient Descent, which updates according to the direction of the gradients, rather than the gradients themselves. In this paper we analyze a stochastic version of NGD and prove its convergence to a global minimum for a wider class of functions: we require the functions to be quasi-convex and locally-Lipschitz. Quasi…
Huawei Technologies Canada（加拿大华为技术有限公司）；University of Alberta（阿尔伯塔大学）
Gradient descent is slow to converge for ill-conditioned problems and non-convex problems. An important technique for acceleration is step-size adaptation. The first part of this paper contains a detailed review of step-size adaptation methods, including Polyak step-size, L4, LossGrad, Adam, IDBD, and Hypergradient descent, and the relation of step-size adaptation to meta-gradient methods. In the second part of this paper, we propose a new class of methods of accelerating gradient descent that have some distinctiveness from existing techniques. The new methods, which we call {\em step-size pla…
We propose new restarting strategies for accelerated gradient and accelerated coordinate descent methods. Our main contribution is to show that the restarted method has a geometric rate of convergence for any restarting frequency, and so it allows us to take profit of restarting even when we do not know the strong convexity coefficient. The scheme can be combined with adaptive restarting, leading to the first provable convergence for adaptive restarting schemes with accelerated gradient methods. Finally, we illustrate the properties of the algorithm on a regularized logistic regression problem…
Many machine learning models require a training procedure based on running stochastic gradient descent. A key element for the efficiency of those algorithms is the choice of the learning rate schedule. While finding good learning rates schedules using Bayesian optimisation has been tackled by several authors, adapting it dynamically in a data-driven way is an open question. This is of high practical importance to users that need to train a single, expensive model. To tackle this problem, we introduce an original probabilistic model for traces of optimisers, based on latent Gaussian processes a…
Learning rate adaptation is a popular topic in machine learning. Gradient Descent trains neural nerwork with a fixed learning rate. Learning rate adaptation is proposed to accelerate the training process through adjusting the step size in the training session. Famous works include Momentum, Adam and Hypergradient. Hypergradient is the most special one. Hypergradient achieved adaptation by calculating the derivative of learning rate with respect to cost function and utilizing gradient descent for learning rate. However, Hypergradient is still not perfect. In practice, Hypergradient fail to decr…
The practical performance of online stochastic gradient descent algorithms is highly dependent on the chosen step size, which must be tediously hand-tuned in many applications. The same is true for more advanced variants of stochastic gradients, such as SAGA, SVRG, or AdaGrad. Here we propose to adapt the step size by performing a gradient descent on the step size itself, viewing the whole performance of the learning trajectory as a function of step size. Importantly, this adaptation can be computed online at little cost, without having to iterate backward passes over the full data.
Max-Planck-Institute for Intelligent Systems（马克斯·普朗克智能系统研究所）
We propose a stepsize adaptation scheme for stochastic gradient descent. It operates directly with the loss function and rescales the gradient in order to make fixed predicted progress on the loss. We demonstrate its capabilities by conclusively improving the performance of Adam and Momentum optimizers. The enhanced optimizers with default hyperparameters consistently outperform their constant stepsize counterparts, even the best ones, without a measurable increase in computational cost. The performance is validated on multiple architectures including dense nets, CNNs, ResNets, and the recurre…
Robert Bosch GmbH（罗伯特·博世有限公司）；University of Freiburg（弗莱堡大学）
Due to their high computational complexity, deep neural networks are still limited to powerful processing units. To promote a reduced model complexity by dint of low-bit fixed-point quantization, we propose a gradient-based optimization strategy to generate a symmetric mixture of Gaussian modes (SGM) where each mode belongs to a particular quantization stage. We achieve 2-bit state-of-the-art performance and illustrate the model's ability for self-dependent weight adaptation during training.
The Ohio State University（俄亥俄州立大学）；Duke University（杜克大学）
Two types of zeroth-order stochastic algorithms have recently been designed for nonconvex optimization respectively based on the first-order techniques SVRG and SARAH/SPIDER. This paper addresses several important issues that are still open in these methods. First, all existing SVRG-type zeroth-order algorithms suffer from worse function query complexities than either zeroth-order gradient descent (ZO-GD) or stochastic gradient descent (ZO-SGD). In this paper, we propose a new algorithm ZO-SVRG-Coord-Rand and develop a new analysis for an existing ZO-SVRG-Coord algorithm proposed in Liu et al.…
Domain adaptation aims at adapting the knowledge acquired on a source domain to a new different but related target domain. Several approaches have beenproposed for classification tasks in the unsupervised scenario, where no labeled target data are available. Most of the attention has been dedicated to searching a new domain-invariant representation, leaving the definition of the prediction function to a second stage. Here we propose to learn both jointly. Specifically we learn the source subspace that best matches the target subspace while at the same time minimizing a regularized misclassific…
In Iterative Learning Control (ILC), a sequence of feedforward control actions is generated at each iteration on the basis of partial model knowledge and past measurements with the goal of steering the system toward a desired reference trajectory. This is framed here as an online learning task, where the decision-maker takes sequential decisions by solving a sequence of optimization problems having only partial knowledge of the cost functions. Having established this connection, the performance of an online gradient-descent based scheme using inexact gradient information is analyzed in the set…
Borealis AI（北极光人工智能）；University of Alberta（阿尔伯塔大学）
There is a long history of using meta learning as representation learning, specifically for determining the relevance of inputs. In this paper, we examine an instance of meta-learning in which feature relevance is learned by adapting step size parameters of stochastic gradient descent---building on a variety of prior work in stochastic approximation, machine learning, and artificial neural networks. In particular, we focus on stochastic meta-descent introduced in the Incremental Delta-Bar-Delta (IDBD) algorithm for setting individual step sizes for each feature of a linear function approximato…
The Ohio State University（俄亥俄州立大学）；Seoul National University（首尔国立大学）
Minimax optimization plays a key role in adversarial training of machine learning algorithms, such as learning generative models, domain adaptation, privacy preservation, and robust learning. In this paper, we demonstrate the failure of alternating gradient descent in minimax optimization problems due to the discontinuity of solutions of the inner maximization. To address this, we propose a new epsilon-subgradient descent algorithm that addresses this problem by simultaneously tracking K candidate solutions. Practically, the algorithm can find solutions that previous saddle-point algorithms ca…
We provide a simple and generic adaptive restart scheme for convex optimization that is able to achieve worst-case bounds matching (up to constant multiplicative factors) optimal restart schemes that require knowledge of problem specific constants. The scheme triggers restarts whenever there is sufficient reduction of a distance-based potential function. This potential function is always computable. We apply the scheme to obtain the first adaptive restart algorithm for saddle-point algorithms including primal-dual hybrid gradient (PDHG) and extragradient. The method improves the worst-case bou…
The monotone variational inequality is a central problem in mathematical programming that unifies and generalizes many important settings such as smooth convex optimization, two-player zero-sum games, convex-concave saddle point problems, etc. The extragradient algorithm by Korpelevich [1976] and the optimistic gradient descent-ascent algorithm by Popov [1980] are arguably the two most classical and popular methods for solving monotone variational inequalities. Despite their long histories, the following major problem remains open. What is the last-iterate convergence rate of the extragradient…
University of Alberta（阿尔伯塔大学）；Google DeepMind（谷歌DeepMind）
This paper investigates different vector step-size adaptation approaches for non-stationary online, continual prediction problems. Vanilla stochastic gradient descent can be considerably improved by scaling the update with a vector of appropriately chosen step-sizes. Many methods, including AdaGrad, RMSProp, and AMSGrad, keep statistics about the learning process to approximate a second order update---a vector approximation of the inverse Hessian. Another family of approaches use meta-gradient descent to adapt the step-size parameters to minimize prediction error. These meta-descent strategies…
University of Washington（华盛顿大学）；Université de Montréal（蒙特利尔大学）；Mila（米拉人工智能研究所）；UC Berkeley（加州大学伯克利分校）
We consider the smooth convex-concave bilinearly-coupled saddle-point problem, $\min_{\mathbf{x}}\max_{\mathbf{y}}~F(\mathbf{x}) + H(\mathbf{x},\mathbf{y}) - G(\mathbf{y})$, where one has access to stochastic first-order oracles for $F$, $G$ as well as the bilinear coupling function $H$. Building upon standard stochastic extragradient analysis for variational inequalities, we present a stochastic \emph{accelerated gradient-extragradient (AG-EG)} descent-ascent algorithm that combines extragradient and Nesterov's acceleration in general stochastic settings. This algorithm leverages scheduled re…
Private machine learning involves addition of noise while training, resulting in lower accuracy. Intuitively, greater stability can imply greater privacy and improve this privacy-utility tradeoff. We study this role of stability in private empirical risk minimization, where differential privacy is achieved by output perturbation, and establish a corresponding theoretical result showing that for strongly-convex loss functions, an algorithm with uniform stability of $β$ implies a bound of $O(\sqrtβ)$ on the scale of noise required for differential privacy. The result applies to both explicit reg…
Department of Engineering Sciences and Applied Mathematics, Northwestern University, Evanston, IL, United States（工程科学与应用数学系，西北大学，埃文斯顿，伊利诺伊州，美国）；Department of Industrial Engineering（工业工程系）；Applied Mathematics, Northwestern University, Evanston, IL, United States（应用数学，西北大学，埃文斯顿，伊利诺伊州，美国）
Optimizers like Adam and AdaGrad have been very successful in training large-scale neural networks. Yet, the performance of these methods is heavily dependent on a carefully tuned learning rate schedule. We show that in many large-scale applications, augmenting a given optimizer with an adaptive tuning method of the step-size greatly improves the performance. More precisely, we maintain a global step-size scale for the update as well as a gain factor for each coordinate. We adjust the global scale based on the alignment of the average gradient and the current gradient vectors. A similar approa…
By analyzing accelerated proximal gradient methods under a local quadratic growth condition, we show that restarting these algorithms at any frequency gives a globally linearly convergent algorithm. This result was previously known only for long enough frequencies. Then, as the rate of convergence depends on the match between the frequency and the quadratic error bound, we design a scheme to automatically adapt the frequency of restart from the observed decrease of the norm of the gradient mapping. Our algorithm has a better theoretical bound than previously proposed methods for the adaptation…
Learning from data streams is an increasingly important topic in data mining, machine learning, and artificial intelligence in general. A major focus in the data stream literature is on designing methods that can deal with concept drift, a challenge where the generating distribution changes over time. A general assumption in most of this literature is that instances are independently distributed in the stream. In this work we show that, in the context of concept drift, this assumption is contradictory, and that the presence of concept drift necessarily implies temporal dependence; and thus som…
Chongqing University of Posts and Telecommunications（重庆邮电大学）；Chongqing Institute of Green and Intelligent Technology, Chinese Academy of Sciences（中国科学院重庆绿色智能技术研究院）；Southwest University（西南大学）
A high-dimensional and incomplete (HDI) matrix can describe the complex interactions among numerous nodes in various big data-related applications. A stochastic gradient descent (SGD)-based latent factor analysis (LFA) model is remarkably effective in extracting valuable information from an HDI matrix. However, such a model commonly encounters the problem of slow convergence because a standard SGD algorithm learns a latent factor relying on the stochastic gradient of current instance error only without considering past update information. To address this critical issue, this paper innovatively…
