Paper removes bounded gradient assumption for SGD in nonconvex learning.
problem Existing theoretical results for SGD in nonconvex learning require uniform boundedness of gradients, which is hard to verify.
method Establishes sufficient conditions for SGD convergence without bounded gradient assumption.
result SGD achieves optimal convergence rates for nonconvex and gradient-dominated objectives.
Paper develops algorithms for nonsmooth, nonconvex statistical learning problems.
problem Nonsmooth and nonconvex objectives in statistical learning.
method Bregman-surrogate algorithm framework, including local linear approximation, mirror descent, iterative thresholding, DC programming.
result Global convergence rates for nonconvex and nonsmooth objectives in high dimensions.
This paper establishes lower bounds for smooth nonconvex finite-sum optimization.
problem Understanding the complexity of finding optimal solutions in nonconvex finite-sum optimization.
method Proving tight lower bounds for the complexity of finding ε-suboptimal points and ε-approximate stationary points.
result Existing algorithms achieve optimal IFO complexity up to logarithmic factors.
MAML optimizes shared priors for subtasks in a nonconvex meta-objective.
problem Understanding global optimality of MAML for nonconvex meta-objectives.
method Characterizes optimality gap of MAML stationary points via first-order optimization methods.
result Establishes global optimality of MAML for both RL and supervised learning.
Safe reinforcement learning with nonconvex constraints using convex approximations.
problem Safe reinforcement learning with nonlinear function approximation.
method Constructing surrogate convex constrained optimization problems by replacing nonconvex functions with convex quadratic functions.
result Solutions to surrogate problems converge to a stationary point of the original nonconvex problem.
New algorithm solves nonconvex-convex minimax problems efficiently.
problem Solving nonconvex-convex minimax problems with nonsmooth, nonconvex, and nonlinearity.
method Hybrid variance-reduced SGD algorithm combining smoothing and biased techniques.
result Achieves O(T^(-2/3)) convergence rate and best oracle complexity.
As surrogate functions of L0-norm, many nonconvex penalty functions have been proposed to enhance the sparse vector recovery. It is easy to extend these nonconvex penalty functions on singular values of a matrix to enhance low-rank matrix recovery. However, different from convex optimization, solving the nonconvex l…
The stochastic gradient (SG) method can minimize an objective function composed of a large number of differentiable functions, or solve a stochastic optimization problem, to a moderate accuracy. The block coordinate descent/update (BCD) method, on the other hand, handles problems with multiple blocks of variables by up…
In this paper, we focus on solving an important class of nonconvex optimization problems which includes many problems for example signal processing over a networked multi-agent system and distributed learning over networks. Motivated by many applications in which the local objective function is the sum of smooth but po…
Continuous optimization is an important problem in many areas of AI, including vision, robotics, probabilistic inference, and machine learning. Unfortunately, most real-world optimization problems are nonconvex, causing standard convex techniques to find only local optima, even with extensions like random restarts and …
Improves manifold learning by identifying and fixing stuck points.
problem Local minima in nonlinear embedding methods lead to poor data structure visualization.
method Introduces a method to temporarily allow pressured points to use an extra dimension in the embedding space.
result Significantly improves the objective function value of existing methods even after getting stuck in a poor local minimum.
We provide novel theoretical results regarding local optima of regularized M-estimators, allowing for nonconvexity in both loss and penalty functions. Under restricted strong convexity on the loss and suitable regularity conditions on the penalty, we prove that \emph{any stationary point} of the composite objective f…
Cubic-regularized Newton's method (CR) is a popular algorithm that guarantees to produce a second-order stationary solution for solving nonconvex optimization problems. However, existing understandings of the convergence rate of CR are conditioned on special types of geometrical properties of the objective function. In…
Distributed algorithm converges to global optima with Gaussian noise.
problem Nonconvex optimization in network-based multi-agent settings.
method Consensus+innovations type algorithm with annealing-type approach and Gaussian noise.
result Convergence to the set of global optima of the sum function.
Paper proposes an algorithm to solve complex minimax problems efficiently.
problem Stochastic nonconvex-concave minimax problems in various fields.
method Accelerated first-order regularized momentum descent ascent algorithm (FORMDA).
result Achieves best-known complexity bound of ildeO(ε−6.5) for single-loop algorithms. Many optimization algorithms converge to stationary points. When the underlying problem is nonconvex, they may get trapped at local minimizers and occasionally stagnate near saddle points. We propose the Run-and-Inspect Method, which adds an "inspect" phase to existing algorithms that helps escape from non-global stati…
Large learning rates lead to various implicit biases in nonconvex optimization.
problem Understanding the conditions under which large learning rates yield edge of stability, balancing, and catapult phenomena.
method Developed a global convergence theory for nonconvex functions without globally Lipschitz continuous gradient, focusing on functions with good regularity.
result These implicit biases are more likely to occur in functions with good regularity, and large learning rates favor flatter regions.
We demonstrate that the primal-dual witness proof method may be used to establish variable selection consistency and ℓ∞-bounds for sparse regression problems, even when the loss function and/or regularizer are nonconvex. Using this method, we derive two theorems concerning support recovery and ℓ∞-…
Improved complexity for smooth nonconvex optimization using quasi-Newton methods.
problem Finding ε-first-order stationary points of smooth functions with gradient information only.
method Two-level online learning approach involving quasi-Newton methods.
result Gradient complexity improved to O(d^(1/4)ε^(-13/8)) for d = O(ε^(-1/2)).
We study finite-sum nonconvex optimization problems, where the objective function is an average of n nonconvex functions. We propose a new stochastic gradient descent algorithm based on nested variance reduction. Compared with conventional stochastic variance reduced gradient (SVRG) algorithm that uses two reference …
We study Frank-Wolfe methods for nonconvex stochastic and finite-sum optimization problems. Frank-Wolfe methods (in the convex case) have gained tremendous recent interest in machine learning and optimization communities due to their projection-free property and their ability to exploit structured constraints. However,…
PPGD solves nonconvex nonsmooth optimization problems without KL property.
problem Nonconvex and nonsmooth optimization problems in statistics and machine learning.
method Projective Proximal Gradient Descent (PPGD) for solving a class of nonconvex and nonsmooth problems.
result PPGD achieves a fast convergence rate of O(1/k^2) for k ≥ k_0.
We analyze stochastic gradient algorithms for optimizing nonconvex, nonsmooth finite-sum problems. In particular, the objective function is given by the summation of a differentiable (possibly nonconvex) component, together with a possibly non-differentiable but convex component. We propose a proximal stochastic gradie…
In this work, we present a globalized stochastic semismooth Newton method for solving stochastic optimization problems involving smooth nonconvex and nonsmooth convex terms in the objective function. We assume that only noisy gradient and Hessian information of the smooth part of the objective function is available via…
CBO interprets as SGD, leading to global convergence for nonconvex functions.
problem Understanding and improving gradient-based learning algorithms.
method Interpreting CBO as a stochastic relaxation of SGD.
result CBO provably converges globally to minimizers for nonsmooth nonconvex functions.
The paper solves video object segmentation without supervision using nonconvex optimization.
problem Unsupervised video object segmentation via background subtraction.
method Formulates the problem as a nonnegative variant of robust principal component analysis, ensuring global optimality under certain conditions.
result Conditions guaranteeing the uniqueness and global optimality of object segmentation are derived and demonstrated with real data.
Adaptive gradient methods are workhorses in deep learning. However, the convergence guarantees of adaptive gradient methods for nonconvex optimization have not been thoroughly studied. In this paper, we provide a fine-grained convergence analysis for a general class of adaptive gradient methods including AMSGrad, RMSPr…
In this paper, we consider the convergence of an abstract inexact nonconvex and nonsmooth algorithm. We promise a pseudo sufficient descent condition and a pseudo relative error condition, which are both related to an auxiliary sequence, for the algorithm; and a continuity condition is assumed to hold. In fact, a lot o…
Laplace kernel feature selection offers statistical guarantees for nonparametric models with few samples.
problem Statistical guarantees for kernel-based feature selection in nonconvex optimization problems.
method Sharp characterization of the gradient of the objective function for Laplace kernel feature selection.
result Model-selection consistency for Laplace kernel-based feature selection in nonparametric settings with n∼logp samples. We study a stochastic and distributed algorithm for nonconvex problems whose objective consists of a sum of N nonconvex Li/N-smooth functions, plus a nonsmooth regularizer. The proposed NonconvEx primal-dual SpliTTing (NESTT) algorithm splits the problem into N subproblems, and utilizes an augmented Lagrangian b…
PWGF escapes saddle points in nonconvex optimization.
problem Escaping saddle points in nonconvex optimization.
method PWGF uses noisy perturbations via Gaussian process to escape saddle points.
result PWGF achieves second-order optimality for nonconvex objectives.
CD methods tackle nonconvex optimization with three terms, achieving critical points.
problem Minimizing nonconvex functions with specific structure.
method Developed randomized CD, randomly permuted CD, and accelerated CD methods.
result CD methods converge to critical points with sublinear complexity.
Within the unmanageably large class of nonconvex optimization, we consider the rich subclass of nonsmooth problems that have composite objectives---this already includes the extensively studied convex, composite objective problems as a special case. For this subclass, we introduce a powerful, new framework that permits…
We propose a general theory for studying the \xl{landscape} of nonconvex \xl{optimization} with underlying symmetric structures \tz{for a class of machine learning problems (e.g., low-rank matrix factorization, phase retrieval, and deep linear neural networks)}. In specific, we characterize the locations of stationary …
Study uncovers complex critical points in tensor decomposition.
problem Nonconvex optimization of symmetric tensor decomposition.
method Utilized symmetry to construct critical points and analyze Hessian.
result Obtained precise analytic estimates on objective function and Hessian.
New method solves nonconvex optimization problems without explicit gradients.
problem Solving nonconvex optimization problems with multiple nonsmooth penalties.
method Zeroth-order stochastic ADMM methods (ZO-SVRG-ADMM and ZO-SAGA-ADMM).
result Convergence rate of O(1/T) for nonconvex optimization. New algorithm solves structured nonconvex-nonconcave min-max problems.
problem Min-max optimization challenges in deep learning.
method Generalized extragradient algorithm for structured nonconvex-nonconcave problems.
result Algorithm converges to stationary points in Euclidean and ℓp spaces. Proposes BMME for optimizing nonsmooth nonconvex problems with block structure.
problem Optimizing nonsmooth nonconvex problems with block structure.
method Block Alternating Bregman Majorization Minimization with Extrapolation (BMME).
result Subsequential convergence to a first-order stationary point under mild assumptions, global convergence under stronger conditions.
New algorithm for nonconvex optimization on constrained Riemannian manifolds converges quickly.
problem Optimization on constrained Riemannian manifolds.
method Block majorization-minimization (BMM) for smooth nonconvex objectives with Riemannian constraints.
result Converges to stationary points within O(ε−2) iterations. Accelerated gradient method tackles nonconvex penalties in sparse learning.
problem Optimizing nonconvex penalties in sparse statistical learning.
method Generalized Nesterov's accelerated gradient method with hyperparameter optimization.
result Convergence can be made considerably faster with optimal hyperparameters.
Two-Timescale EM Methods improve EM for nonconvex models.
problem Nonconvex latent variable models are challenging for EM.
method Two-stage stochastic updates to handle nonconvex optimization.
result Global convergence for nonconvex objective functions.
Paper proposes algorithms for solving nonconvex-nonconcave problems with complexity guarantees.
problem Nonconvex-nonconcave minimax problems with PL condition.
method Zeroth-order AGDA and VRAGDA algorithms.
result Iteration complexities for obtaining ε-stationary points.
Paper tackles Hessian/Jacobian-free stochastic bilevel optimization with O(ε−1.5) complexity.
problem Nonconvex-strongly-convex bilevel optimization problem.
method FdeHBO optimizer with finite-difference Hessian/Jacobian-vector approximation and momentum.
result FdeHBO achieves O(ε−1.5) iterations for ε-accurate stationary point. Stabilized SVRG finds second-order stationary points efficiently in nonconvex optimization.
problem Finding second-order stationary points in nonconvex optimization.
method Stabilized SVRG, a variant of SVRG.
result Stabilized SVRG finds an ε-second-order stationary point using only O(n^2/3/ε^2 + n/ε^1.5) stochastic gradients.
Paper develops methods for statistical inference with SGD in nonconvex optimization.
problem Statistical inference for nonconvex optimization problems.
method Proposes two online inferential procedures combining SGD and bootstrap techniques.
result Establishes error convergence rates and asymptotically valid bootstrap confidence intervals.
Method solves nonconvex constrained optimization problems with a new augmented Lagrangian approach.
problem Nonconvex composite functional constraints with inequality constraints.
method First-order augmented Lagrangian method with smoothed prox-linear reformulation.
result Explicit convergence rates for the proposed method in terms of KKT residual.
SONATA algorithm converges to solutions of nonconvex smooth functions with KL property.
problem Decentralized optimization over networks with nonconvex smooth functions and convex constraints.
method Decentralized gradient-tracking algorithm SONATA under the KL property.
result SONATA converges to stationary solutions at R-linear rate for θ∈(0,1/2], sublinear rate for θ∈(1/2,1), and R-linear rate for θ=0. Unified Lagrangian-based methods for nonsmooth nonconvex optimization.
problem Minimizing nonsmooth nonconvex functions with constraints.
method Developed a unified framework for Lagrangian-based methods using subgradient updates.
result Global convergence guarantees for the proposed framework under mild conditions.