arXiv research

New method bounds stochastic subgradient methods with heavy-tailed noise.

The paper guarantees global stability for stochastic subgradient methods in nonsmooth nonconvex optimization.

A distributed subgradient method tackles non-convex optimization problems in networks.

We develop model-based methods for solving stochastic convex optimization problems, introducing the approximate-proximal point, or aProx, family, which includes stochastic subgradient, proximal point, and bundle methods. When the modeling approaches we propose are appropriately accurate, the methods enjoy stronger conv…

We describe novel subgradient methods for a broad class of matrix optimization problems involving nuclear norm regularization. Unlike existing approaches, our method executes very cheap iterations by combining low-rank stochastic subgradients with efficient incremental SVD updates, made possible by highly optimized and…

We show that the Subgradient algorithm is universal for online learning on the simplex in the sense that it simultaneously achieves $O(\sqrt N)$ regret for adversarial costs and $O(1)$ pseudo-regret for i.i.d costs. To the best of our knowledge this is the first demonstration of a universal algorithm on the simplex tha…

In this note, we present a new averaging technique for the projected stochastic subgradient method. By using a weighted average with a weight of t+1 for each iterate w_t at iteration t, we obtain the convergence rate of O(1/t) with both an easy proof and an easy implementation. The new scheme is compared empirically to…

Bayesian max-margin models have shown superiority in various practical applications, such as text categorization, collaborative prediction, social network link prediction and crowdsourcing, and they conjoin the flexibility of Bayesian modeling and predictive strengths of max-margin learning. However, Monte Carlo sampli…

Stochastic subgradient descent avoids critical points in definable functions.

In this paper, a new theory is developed for first-order stochastic convex optimization, showing that the global convergence rate is sufficiently quantified by a local growth rate of the objective function in a neighborhood of the optimal solutions. In particular, if the objective function $F(\mathbf w)$ in the $ε$-sub…

This work establishes uniform convergence of subdifferentials in stochastic optimization.

New algorithms optimize spectral risk measures, improving interpolation between average and worst-case performance.

Paper presents an efficient algorithm for learning minimax risk classifiers with large-scale data.

We consider the problem of minimizing a convex risk with stochastic subgradients guaranteeing $ε$-locally differentially private ($ε$-LDP). While it has been shown that stochastic optimization is possible with $ε$-LDP via the standard SGD (Song et al., 2013), its convergence rate largely depends on the learning rate, w…

We consider the problem of unconstrained online convex optimization (OCO) with sub-exponential noise, a strictly more general problem than the standard OCO. In this setting, the learner receives a subgradient of the loss functions corrupted by sub-exponential noise and strives to achieve optimal regret guarantee, witho…

Optimized method tackles convex optimization with heavy-tailed noise.

Unified Lagrangian-based methods for nonsmooth nonconvex optimization.

New algorithms accelerate model-based optimization for stochastic problems.

We propose a randomized block-coordinate variant of the classic Frank-Wolfe algorithm for convex optimization with block-separable constraints. Despite its lower iteration cost, we show that it achieves a similar convergence rate in duality gap as the full Frank-Wolfe algorithm. We also show that, when applied to the d…

Stochastic algorithm achieves sublinear convergence for bi-objective optimization.

Study on Adam-family methods for nonsmooth optimization with convergence guarantees.

SGD avoids critical points on weakly convex functions.

Paper addresses Byzantine attacks in decentralized optimization over networks.

Nesterov's extrapolation improves convergence in nonsmooth optimization.

We propose graph-dependent implicit regularisation strategies for distributed stochastic subgradient descent (Distributed SGD) for convex problems in multi-agent learning. Under the standard assumptions of convexity, Lipschitz continuity, and smoothness, we establish statistical learning rates that retain, up to logari…

SGD converges to critical points of normalized margin in late-stage training for homogeneous neural networks.

In this work we consider the stochastic minimization of nonsmooth convex loss functions, a central problem in machine learning. We propose a novel algorithm called Accelerated Nonsmooth Stochastic Gradient Descent (ANSGD), which exploits the structure of common nonsmooth loss functions to achieve optimal convergence ra…

Improved subgradient method tackles ill-conditioned composite optimization problems.

We generalize stochastic subgradient descent methods to situations in which we do not receive independent samples from the distribution over which we optimize, but instead receive samples that are coupled over time. We show that as long as the source of randomness is suitably ergodic---it converges quickly enough to a …

Investor optimizes utility in a market with endogenous pricing.

Optimizes distributions robustly with Sinkhorn distance.

New Max-Plus neural network exploits subgradient sparsity for efficient training.

The paper derives subgradient estimates for a specific nonlinear subparabolic equation on pseudo-Hermitian manifolds.

Despite remarkable empirical success, the training dynamics of generative adversarial networks (GAN), which involves solving a minimax game using stochastic gradients, is still poorly understood. In this work, we analyze last-iterate convergence of simultaneous gradient descent (simGD) and its variants under the assump…

The analysis in Part I revealed interesting properties for subgradient learning algorithms in the context of stochastic optimization when gradient noise is present. These algorithms are used when the risk functions are non-smooth and involve non-differentiable components. They have been long recognized as being slow co…

Proof of convergence for multi-objective optimization using inverse reinforcement learning.

We study computational and statistical consequences of problem geometry in stochastic and online optimization. By focusing on constraint set and gradient geometry, we characterize the problem families for which stochastic- and adaptive-gradient methods are (minimax) optimal and, conversely, when nonlinear updates -- su…

We consider the stochastic nested composition optimization problem where the objective is a composition of two expected-value functions. We proposed the stochastic ADMM to solve this complicated objective. In order to find an $ε$ stationary point where the expected norm of the subgradient of corresponding augmented Lag…

Inexact subgradient methods work well for semialgebraic functions with additive errors.

We consider in this paper a class of composite optimization problems whose objective function is given by the summation of a general smooth and nonsmooth component, together with a relatively simple nonsmooth term. We present a new class of first-order methods, namely the gradient sliding algorithms, which can skip the…

New algorithms achieve high-probability parameter-free regret in online convex optimization with heavy-tailed data.

Study proves convergence of subgradients for optimal transport-based objectives.

The paper accelerates ISTA and FISTA algorithms for composite optimization problems.

New adaptive methods solve weakly convex stochastic optimization problems.

The paper relaxes assumptions for analyzing stochastic optimization algorithms.

In this paper we study integer multiplicity rectifiable currents carried by the subgradient (subdifferential) graphs of semi-convex functions on a $n$-dimensional convex domain, and show a weak continuity theorem with respect to pointwise convergence for such currents. As an application, the $k$-Hessian measures are ca…

New algorithms solve large-scale convex regression problems.

Study robust recovery of low-rank matrices from corrupted measurements without rank prior.