arXiv research

The paper provides Gaussian approximations for decentralized Federated Learning.

Smoothed SGD improves quantile estimation without crossing curves.

The paper analyzes the variance of different shuffling methods in stochastic gradient descent.

The paper clarifies the approximation of SGD with Ito SDEs for finite learning rates.

Uniform diffusion approximation for SGD in non-convex settings.

Simplified SGD interpretation as Ito process for broader applicability.

Paper improves confidence set construction for SGD using multiplier bootstrap.

The paper investigates heavy-tailed behavior in offline SGD, showing it approximates power-law tails.

Stochastic Gradient Descent with a constant learning rate (constant SGD) simulates a Markov chain with a stationary distribution. With this perspective, we derive several new results. (1) We show that constant SGD can be used as an approximate Bayesian posterior inference algorithm. Specifically, we show how to adjust …

HA-SME models SGD dynamics with Hessian info for better escaping behaviors.

We study the Stochastic Gradient Descent (SGD) method in nonconvex optimization problems from the point of view of approximating diffusion processes. We prove rigorously that the diffusion process can approximate the SGD algorithm weakly using the weak form of master equation for probability evolution. In the small ste…

The paper analyzes SGD and its continuous counterpart, improving convergence rates and approximation results.

We revisit the choice of SGD for training deep neural networks by reconsidering the appropriate geometry in which to optimize the weights. We argue for a geometry invariant to rescaling of weights that does not affect the output of the network, and suggest Path-SGD, which is an approximate steepest descent method with …

Heavy-tailed distributions emerge in SGD's parameter evolution.

We design a stochastic algorithm to train any smooth neural network to $\varepsilon$-approximate local minima, using $O(\varepsilon^{-3.25})$ backpropagations. The best result was essentially $O(\varepsilon^{-4})$ by SGD. More broadly, it finds $\varepsilon$-approximate local minima of any smooth nonconvex function in …

New bounds for M-SGD show its error distribution is nearly Gaussian.

Study on the noise in SGD minibatches near local minima.

This work characterizes the benefits of averaging schemes widely used in conjunction with stochastic gradient descent (SGD). In particular, this work provides a sharp analysis of: (1) mini-batching, a method of averaging many samples of a stochastic gradient to both reduce the variance of the stochastic gradient estima…

This work improves SGD convergence by adaptively adjusting batch sizes.

New algorithm resists Byzantine attacks in distributed SGD for heterogeneous data.

Second-order optimization methods such as natural gradient descent have the potential to speed up training of neural networks by correcting for the curvature of the loss function. Unfortunately, the exact natural gradient is impractical to compute for large models, and most approximations either require an expensive it…

Three DP variants linked, improving SGD privacy bounds.

Stochastic gradient descent (SGD) has been found to be surprisingly effective in training a variety of deep neural networks. However, there is still a lack of understanding on how and why SGD can train these complex networks towards a global minimum. In this study, we establish the convergence of SGD to a global minimu…

Study reveals convergence properties of SGD with random learning rate.

Classical stochastic gradient methods for optimization rely on noisy gradient approximations that become progressively less accurate as iterates approach a solution. The large noise and small signal in the resulting gradients makes it difficult to use them for adaptive stepsize selection and automatic stopping. We prop…

Machine learning, especially deep neural networks, has been rapidly developed in fields including computer vision, speech recognition and reinforcement learning. Although Mini-batch SGD is one of the most popular stochastic optimization methods in training deep networks, it shows a slow convergence rate due to the larg…

This work improves privacy-generalization bounds for DP-SGD.

Stochastic Gradient Descent (SGD) is an important algorithm in machine learning. With constant learning rates, it is a stochastic process that, after an initial phase of convergence, generates samples from a stationary distribution. We show that SGD with constant rates can be effectively used as an approximate posterio…

Stochastic optimization improves semi-discrete OT map estimation with a minimax rate.

We develop the mathematical foundations of the stochastic modified equations (SME) framework for analyzing the dynamics of stochastic gradient algorithms, where the latter is approximated by a class of stochastic differential equations with small noise parameters. We prove that this approximation can be understood math…

Cheap methods improve uncertainty in SGD solutions.

New approach proves convergence of SA and SGD with weaker conditions.

Paper develops methods for statistical inference with SGD in nonconvex optimization.

Reinforcement Learning (RL) algorithms allow artificial agents to improve their action selections so as to increase rewarding experiences in their environments. Deep Reinforcement Learning algorithms require solving a nonconvex and nonlinear unconstrained optimization problem. Methods for solving the optimization probl…

This work analyzes SGD for SGMs, providing convergence rates and error bounds.

New perspective on SGD reveals short-range memory effects in deep learning.

Stochastic gradient descent (SGD) is a well known method for regression and classification tasks. However, it is an inherently sequential algorithm at each step, the processing of the current example depends on the parameters learned from the previous examples. Prior approaches to parallelizing linear learners using SG…

Deep neural networks are typically trained by optimizing a loss function with an SGD variant, in conjunction with a decaying learning rate, until convergence. We show that simple averaging of multiple points along the trajectory of SGD, with a cyclical or constant learning rate, leads to better generalization than conv…

Study of SGD with state-dependent noise, improving escape from local minima.

We investigate the dynamical and convergent properties of stochastic gradient descent (SGD) applied to Deep Neural Networks (DNNs). Characterizing the relation between learning rate, batch size and the properties of the final minima, such as width or generalization, remains an open question. In order to tackle this pro…

Top-k sparsification reduces deep learning communication costs.

We provide non-asymptotic convergence rates of the Polyak-Ruppert averaged stochastic gradient descent (SGD) to a normal random vector for a class of twice-differentiable test functions. A crucial intermediate step is proving a non-asymptotic martingale central limit theorem (CLT), i.e., establishing the rates of conve…

Stochastic gradient descent (SGD) gives an optimal convergence rate when minimizing convex stochastic objectives $f(x)$. However, in terms of making the gradients small, the original SGD does not give an optimal rate, even when $f(x)$ is convex. If $f(x)$ is convex, to find a point with gradient norm $\varepsilon$, we …

This work improves SGD minibatch sampling using determinantal point processes based on orthogonal polynomials.

Study SGD dynamics in high-dimensional models, revealing consistent behavior across different batch sizes and learning rates.

Two-layer networks learn hard GLMs with SGD in high dimensions.

We propose SWA-Gaussian (SWAG), a simple, scalable, and general purpose approach for uncertainty representation and calibration in deep learning. Stochastic Weight Averaging (SWA), which computes the first moment of stochastic gradient descent (SGD) iterates with a modified learning rate schedule, has recently been sho…

The paper analyzes convergence rates for SGD and SHB methods.