We derive a sound positive semi-definite approximation of the Hessian of deep models for which Hessian-vector products are easily computable. This enables us to provide an adaptive SGD learning rate strategy based on the minimization of the local quadratic approximation, which requires just twice the computation of a single SGD run, but performs comparably with grid search on SGD learning rates on different model architectures (CNN with and without residual connections) on classification tasks. We also compare the novel approximation with the Gauss-Newton approximation.
Department of Electrical Engineering, Stanford University（斯坦福大学电气工程系）
We propose a randomized algorithm with quadratic convergence rate for convex optimization problems with a self-concordant, composite, strongly convex objective function. Our method is based on performing an approximate Newton step using a random projection of the Hessian. Our first contribution is to show that, at each iteration, the embedding dimension (or sketch size) can be as small as the effective dimension of the Hessian matrix. Leveraging this novel fundamental result, we design an algorithm with a sketch size proportional to the effective dimension and which exhibits a quadratic rate o…
Information Sciences Institute（信息科学研究所）；University of Southern California（南加州大学）
In this paper, we introduce Apollo, a quasi-Newton method for nonconvex stochastic optimization, which dynamically incorporates the curvature of the loss function by approximating the Hessian via a diagonal matrix. Importantly, the update and storage of the diagonal approximation of Hessian is as efficient as adaptive first-order optimization methods with linear complexity for both time and memory. To handle nonconvexity, we replace the Hessian with its rectified absolute value, which is guaranteed to be positive-definite. Experiments on three tasks of vision and language show that Apollo achi…
First-ish Order Methods: Hessian-aware Scalings of Gradient Descent
School of Mathematics and Physics（数学与物理学院）；University of Queensland（昆士兰大学）；ARC Training Centre for Information Resilience（信息韧性培训中心）；Computer Sciences Department（计算机科学系）；University of Wisconsin-Madison（威斯康星大学麦迪逊分校）
While stochastic gradient descent (SGD) and variants have been surprisingly successful for training deep nets, several aspects of the optimization dynamics and generalization are still not well understood. In this paper, we present new empirical observations and theoretical results on both the optimization dynamics and generalization behavior of SGD for deep nets based on the Hessian of the training loss and associated quantities. We consider three specific research questions: (1) what is the relationship between the Hessian of the loss and the second moment of stochastic gradients (SGs)? (2) …
AI Theory Lab（AI理论实验室）；Huawei（华为）；Machine Learning Research Group and Oxford-Man Institute for Quantitative Finance（机器学习研究组与牛津-曼量化金融研究所）；University of Oxford（牛津大学）
We study the effect of mini-batching on the loss landscape of deep neural networks using spiked, field-dependent random matrix theory. We demonstrate that the magnitude of the extremal values of the batch Hessian are larger than those of the empirical Hessian. We also derive similar results for the Generalised Gauss-Newton matrix approximation of the Hessian. As a consequence of our theorems we derive an analytical expressions for the maximal learning rates as a function of batch size, informing practical training regimens for both stochastic gradient descent (linear scaling) and adaptive algo…
Learning RBMs using standard algorithms such as CD(k) involves gradient descent on the negative log-likelihood. One of the terms in the gradient, which involves expectation w.r.t. the model distribution, is intractable and is obtained through an MCMC estimate. In this work we show that the Hessian of the log-likelihood can be written in terms of covariances of hidden and visible units and hence, all elements of the Hessian can also be estimated using the same MCMC samples with small extra computational costs. Since inverting the Hessian may be computationally expensive, we propose an algorithm…
National Institute of Technology, Agartala（阿加尔塔拉国家理工学院）；Chitkara University Institute of Engineering & Technology, Chitkara University（奇特卡拉大学工程与技术学院）；Mansoura University（曼苏拉大学）
In quantum computing, the variational quantum algorithms (VQAs) are well suited for finding optimal combinations of things in specific applications ranging from chemistry all the way to finance. The training of VQAs with gradient descent optimization algorithm has shown a good convergence. At an early stage, the simulation of variational quantum circuits on noisy intermediate-scale quantum (NISQ) devices suffers from noisy outputs. Just like classical deep learning, it also suffers from vanishing gradient problems. It is a realistic goal to study the topology of loss landscape, to visualize th…
Adaptive pruning-based Newton's method for distributed learning
College of Computer Science and Technology, Ocean University of China（中国海洋大学计算机科学与技术学院）；School of Computer Science and Technology, Shandong University（山东大学计算机科学与技术学院）；School of Electrical Engineering and Computer Science, TU Berlin（柏林工业大学电气工程与计算机科学学院）；Department of Computer Science, Georgia State University（佐治亚州立大学计算机科学系）
Hessian based measures of flatness, such as the trace, Frobenius and spectral norms, have been argued, used and shown to relate to generalisation. In this paper we demonstrate that for feed forward neural networks under the cross entropy loss, we would expect low loss solutions with large weights to have small Hessian based measures of flatness. This implies that solutions obtained using $L2$ regularisation should in principle be sharper than those without, despite generalising better. We show this to be true for logistic regression, multi-layer perceptrons, simple convolutional, pre-activated…
We develop a new algorithm for non-convex stochastic optimization that finds an $ε$-critical point in the optimal $O(ε^{-3})$ stochastic gradient and Hessian-vector product computations. Our algorithm uses Hessian-vector products to "correct" a bias term in the momentum of SGD with momentum. This leads to better gradient estimates in a manner analogous to variance reduction methods. In contrast to prior work, we do not require excessively large batch sizes, and are able to provide an adaptive algorithm whose convergence rate automatically improves with decreasing variance in the gradient estim…
CEREMADE, Université Paris Dauphine - PSL Research University（CEREMADE，巴黎多芬大学-巴黎文理研究大学）
The minimization of the loss function is of paramount importance in deep neural networks. On the other hand, many popular optimization algorithms have been shown to correspond to some evolution equation of gradient flow type. Inspired by the numerical schemes used for general evolution equations we introduce a second order stochastic Runge Kutta method and show that it yields a consistent procedure for the minimization of the loss function. In addition it can be coupled, in an adaptive framework, with a Stochastic Gradient Descent (SGD) to adjust automatically the learning rate of the SGD, wit…
Adaptive optimization methods are well known to achieve superior convergence relative to vanilla gradient methods. The traditional viewpoint in optimization, particularly in convex optimization, explains this improved performance by arguing that, unlike vanilla gradient schemes, adaptive algorithms mimic the behavior of a second-order method by adapting to the global geometry of the loss function. We argue that in the context of neural network optimization, this traditional viewpoint is insufficient. Instead, we advocate for a local trajectory analysis. For iterate trajectories produced by run…
This paper revisits the so-called vanishing gradient phenomenon, which commonly occurs in deep randomly initialized neural networks. Leveraging an in-depth analysis of neural chains, we first show that vanishing gradients cannot be circumvented when the network width scales with less than O(depth), even when initialized with the popular Xavier and He initializations. Second, we extend the analysis to second-order derivatives and show that random i.i.d. initialization also gives rise to Hessian matrices with eigenspectra that vanish as networks grow in depth. Whenever this happens, optimizers a…
We consider least-squares problems with quadratic regularization and propose novel sketching-based iterative methods with an adaptive sketch size. The sketch size can be as small as the effective dimension of the data matrix to guarantee linear convergence. However, a major difficulty in choosing the sketch size in terms of the effective dimension lies in the fact that the latter is usually unknown in practice. Current sketching-based solvers for regularized least-squares fall short on addressing this issue. Our main contribution is to propose adaptive versions of standard sketching-based iter…
TRSVR: An Adaptive Stochastic Trust-Region Method with Variance Reduction
Department of Mathematics, University of California, Berkeley, CA, USA（加州大学伯克利分校数学系）；Boston University, MA, USA（波士顿大学）；Department of IEOR, University of California, Berkeley, CA, USA（加州大学伯克利分校工业工程与运筹学系）
Iterative Hessian sketch (IHS) is an effective sketching method for modeling large-scale data. It was originally proposed by Pilanci and Wainwright (2016; JMLR) based on randomized sketching matrices. However, it is computationally intensive due to the iterative sketch process. In this paper, we analyze the IHS algorithm under the unconstrained least squares problem setting, then propose a deterministic approach for improving IHS via A-optimal subsampling. Our contributions are three-fold: (1) a good initial estimator based on the A-optimal design is suggested; (2) a novel ridged preconditione…
Courant Institute of Mathematical Sciences（库朗数学科学研究所）；New York University（纽约大学）
Recent work has established an empirically successful framework for adapting learning rates for stochastic gradient descent (SGD). This effectively removes all needs for tuning, while automatically reducing learning rates over time on stationary problems, and permitting learning rates to grow appropriately in non-stationary tasks. Here, we extend the idea in three directions, addressing proper minibatch parallelization, including reweighted updates for sparse or orthogonal gradients, improving robustness on non-smooth loss functions, in the process replacing the diagonal Hessian estimation pro…
Bayesian approaches have been successfully integrated into training deep neural networks. One popular family is stochastic gradient Markov chain Monte Carlo methods (SG-MCMC), which have gained increasing interest due to their scalability to handle large datasets and the ability to avoid overfitting. Although standard SG-MCMC methods have shown great performance in a variety of problems, they may be inefficient when the random variables in the target posterior densities have scale differences or are highly correlated. In this work, we present an adaptive Hessian approximated stochastic gradien…
We consider large scale empirical risk minimization (ERM) problems, where both the problem dimension and variable size is large. In these cases, most second order methods are infeasible due to the high cost in both computing the Hessian over all samples and computing its inverse in high dimensions. In this paper, we propose a novel adaptive sample size second-order method, which reduces the cost of computing the Hessian by solving a sequence of ERM problems corresponding to a subset of samples and lowers the cost of computing the Hessian inverse using a truncated eigenvalue decomposition. We s…
University of Maryland（马里兰大学）；Indian Institute of Science（印度科学学院）
We present novel algorithms for simulation optimization using random directions stochastic approximation (RDSA). These include first-order (gradient) as well as second-order (Newton) schemes. We incorporate both continuous-valued as well as discrete-valued perturbations into both our algorithms. The former are chosen to be independent and identically distributed (i.i.d.) symmetric, uniformly distributed random variables (r.v.), while the latter are i.i.d., asymmetric, Bernoulli r.v.s. Our Newton algorithm, with a novel Hessian estimation scheme, requires N-dimensional perturbations and three l…
University of California, Berkeley（加州大学伯克利分校）；NERSC, Lawrence Berkeley National Laboratory（劳伦斯伯克利国家实验室）
We introduce ADAHESSIAN, a second order stochastic optimization algorithm which dynamically incorporates the curvature of the loss function via ADAptive estimates of the HESSIAN. Second order algorithms are among the most powerful optimization algorithms with superior convergence properties as compared to first order methods such as SGD and Adam. The main disadvantage of traditional second order methods is their heavier per iteration computation and poor accuracy as compared to first order methods. To address these, we incorporate several novel approaches in ADAHESSIAN, including: (i) a fast H…
Adaptive learning rate algorithms such as RMSProp are widely used for training deep neural networks. RMSProp offers efficient training since it uses first order gradients to approximate Hessian-based preconditioning. However, since the first order gradients include noise caused by stochastic optimization, the approximation may be inaccurate. In this paper, we propose a novel adaptive learning rate algorithm called SDProp. Its key idea is effective handling of the noise by preconditioning based on covariance matrix. For various neural networks, our approach is more efficient and effective than …
Hessian Aware Low-Rank Perturbation for Order-Robust Continual Learning
University of Western Ontario（西安大略大学）；Vector Institute（矢量研究院）；Beihang University（北京航空航天大学）；Mcgill University（麦吉尔大学）；Laval University（拉瓦尔大学）；Mila（米拉研究所）
University of Pittsburgh（匹兹堡大学）；Simon Fraser University（西蒙弗雷泽大学）；JD Finance America Corporation（京东金融美国公司）
In the paper, we propose a class of efficient momentum-based policy gradient methods for the model-free reinforcement learning, which use adaptive learning rates and do not require any large batches. Specifically, we propose a fast important-sampling momentum-based policy gradient (IS-MBPG) method based on a new momentum-based variance reduced technique and the importance sampling technique. We also propose a fast Hessian-aided momentum-based policy gradient (HA-MBPG) method based on the momentum-based variance reduced technique and the Hessian-aided technique. Moreover, we prove that both the…
CAMP-HiVe: Cyclic Pair Merging based Efficient DNN Pruning with Hessian-Vector Approximation for Resource-Constrained Systems
Engineering University of Louisville（工程大学路易斯维尔大学）；School of Computing University of Georgia（计算学院佐治亚大学）；Department of Computer Science（1 计算机科学系）；Engineering, University of Louisville, KY, USA（工程，路易斯维尔大学，美国肯塔基州）；School of Computing, University of Georgia, Athens, GA, USA（2 计算学院，佐治亚大学，美国亚特兰大州）
