arXiv research

New unbiased gradient estimators for complex optimization problems.

Policy gradient aims to maximize expected return using gradient ascent.

Paper proposes an unbiased optimization method for Bayesian experimental design.

Analytical method approximates ELBO gradient in clutter problem.

Gradient noise improves privacy-protected optimization performance.

This paper improves an online learning algorithm using second-order information.

We propose expected policy gradients (EPG), which unify stochastic policy gradients (SPG) and deterministic policy gradients (DPG) for reinforcement learning. Inspired by expected sarsa, EPG integrates across the action when estimating the gradient, instead of relying only on the action in the sampled trajectory. We es…

Maxout networks study gradients and propose initialization strategies.

Develops methods to estimate gradient of EIG for Bayesian Experimental Design.

Stochastic gradient methods can converge in expectation under heavy-tailed noise.

We consider stochastic gradient descent and its averaging variant for binary classification problems in a reproducing kernel Hilbert space. In the traditional analysis using a consistency property of loss functions, it is known that the expected classification error converges more slowly than the expected risk even whe…

New methods for estimating gradients of expectations using pairwise interactions.

ARSM estimator improves gradient backpropagation for categorical variables.

Improved deep learning models using new attribution priors and expected gradients.

We introduce local expectation gradients which is a general purpose stochastic variational inference algorithm for constructing stochastic gradients through sampling from the variational distribution. This algorithm divides the problem of estimating the stochastic gradients over multiple variational parameters into sma…

A new gradient method reduces variance for non-reparameterizable distributions.

Continuous-time SGD converges under certain conditions, useful for deep learning.

We propose expected policy gradients (EPG), which unify stochastic policy gradients (SPG) and deterministic policy gradients (DPG) for reinforcement learning. Inspired by expected sarsa, EPG integrates (or sums) across actions when estimating the gradient, instead of relying only on the action in the sampled trajectory…

This work bounds the run-time of nonconvex optimization with early stopping.

Gradient-free framework for Bayesian experimental design in complex systems.

A new parallel BO method with exact gradients for multi-objective optimization.

New method for optimizing complex composite functions with reduced variance.

Unified approach for Bayesian optimal experiment design using stochastic gradients.

Vanishing gradients hinder reinforcement finetuning of language models.

Gradient penalty improves GAN performance by inducing a large-margin classifier.

A new unbiased Hessian estimator for expectation-based objectives.

Classical stochastic gradient methods are well suited for minimizing expected-value objective functions. However, they do not apply to the minimization of a nonlinear function involving expected values or a composition of two expected-value functions, i.e., problems of the form $\min_x \mathbf{E}_v [f_v\big(\mathbf{E}_…

Paper develops NPG for risk-averse RL with ECRMs, proving global convergence.

PS-IG improves feature attribution by reducing noise and variance.

This research improves forecasting and testing of risk contributions using Expected Shortfall.

We study gradient Ricci solitons with maximal symmetry. First we show that there are no non-trivial homogeneous gradient Ricci solitons. Thus the most symmetry one can expect is an isometric cohomogeneity one group action. Many examples of cohomogeneity one gradient solitons have been constructed. However, we apply the…

New DP EM algorithm with statistical guarantees for mixture models.

New learning algorithm for real analytic functions without gradient descent.

DG improves policy gradients by weighting actions with a sigmoid of advantage and surprisal.

New framework improves EM algorithm convergence under log-Sobolev inequality.

New methods optimize functions faster with less gradient accuracy needed.

Paper develops probabilistic bounds for a stochastic gradient algorithm in non-convex problems.

This work deals with parallel optimization of expensive objective functions which are modeled as sample realizations of Gaussian processes. The study is formalized as a Bayesian optimization problem, or continuous multi-armed bandit problem, where a batch of q > 0 arms is pulled in parallel at each iteration. Several a…

A new tamed stochastic gradient Hamiltonian Monte Carlo algorithm for superlinearly growing stochastic gradients.

New method reduces complexity for nonconvex optimization problems.

New convergence guarantees for SGDA and SCO under expected co-coercivity.

Paper extends option-critic architecture to estimate natural gradient for reinforcement learning.

New method for unbiased regression reduces excess risk.

We introduce a doubly stochastic proximal gradient algorithm for optimizing a finite average of smooth convex functions, whose gradients depend on numerically expensive expectations. Our main motivation is the acceleration of the optimization of the regularized Cox partial-likelihood (the core model used in survival an…

A new method improves training speed and robustness for deep learning models.

It has been observed in a variety of contexts that gradient descent methods have great success in solving low-rank matrix factorization problems, despite the relevant problem formulation being non-convex. We tackle a particular instance of this scenario, where we seek the $d$-dimensional subspace spanned by a streaming…

Efficiently simulates risk budgeting portfolios using novel algorithms.

Unified framework for training SNNs using EP, faster convergence.