arXiv research

Gradient descent implicitly regularizes neural networks by penalizing large loss gradients.

Gradient model for memristive systems in neurophysiology and neuromorphic circuits.

Paper introduces ICGNs to model convex gradients.

Reinforcement learning algorithms such as the deep deterministic policy gradient algorithm (DDPG) has been widely used in continuous control tasks. However, the model-free DDPG algorithm suffers from high sample complexity. In this paper we consider the deterministic value gradients to improve the sample efficiency of …

Optimal defenses protect FL models from gradient reconstruction attacks.

Gradient sparsification enhances privacy-preserving machine learning models.

A method to improve gradient boosting models using stacking.

Gradient rollback explains neural models by identifying influential training examples.

This research tackles uncertainty in gradient boosting models using ensemble methods.

New research shows input-gradients can be manipulated without changing model's core function, challenging their use for model interpretation.

Proposes VSGD optimizer combining probabilistic and gradient-based methods.

MAGE optimizes policies using action gradients from model-based learning.

A new estimator improves training of probabilistic models with latent Gaussian variables.

Variational inference approximates the posterior distribution of a probabilistic model with a parameterized density by maximizing a lower bound for the model evidence. Modern solutions fit a flexible approximation with stochastic gradient descent, using Monte Carlo approximation for the gradients. This enables variatio…

Gradient amplification boosts deep learning model performance without increasing training time.

A scalable method for BED with implicit models using approximate gradients.

Bayesian inference plays an important role in advancing machine learning, but faces computational challenges when applied to complex models such as deep neural networks. Variational inference circumvents these challenges by formulating Bayesian inference as an optimization problem and solving it using gradient-based op…

Paper proposes an $ε$-policy gradient for online pricing, reducing regret to $\mathcal{O}(\sqrt{T})$.

SGD reduces test error by decorrelating updates.

Gradient-enhanced deep GPs improve multifidelity model accuracy.

Proposes using gradients as features for efficient deep learning adaptation.

Paper develops unbiased gradient estimator for continuous-time models.

Gradient descent dynamics studied for DEQs in linear and single-index models.

New findings on optimal transport gradient for generative models, addressing numerical instabilities.

We introduce a simple algorithm, True Asymptotic Natural Gradient Optimization (TANGO), that converges to a true natural gradient descent in the limit of small learning rates, without explicit Fisher matrix estimation. For quadratic models the algorithm is also an instance of averaged stochastic gradient, where the par…

Adaptive batch sizes improve local gradient methods in distributed training.

Variational inference transforms posterior inference into parametric optimization thereby enabling the use of latent variable models where otherwise impractical. However, variational inference can be finicky when different variational parameters control variables that are strongly correlated under the model. Traditiona…

The natural gradient method has been used effectively in conjugate Gaussian process models, but the non-conjugate case has been largely unexplored. We examine how natural gradients can be used in non-conjugate stochastic settings, together with hyperparameter learning. We conclude that the natural gradient can signific…

In this work we revisit gradient regularization for adversarial robustness with some new ingredients. First, we derive new per-image theoretical robustness bounds based on local gradient information. These bounds strongly motivate input gradient regularization. Second, we implement a scaleable version of input gradient…

Recent advances in policy gradient methods and deep learning have demonstrated their applicability for complex reinforcement learning problems. However, the variance of the performance gradient estimates obtained from the simulation is often excessive, leading to poor sample efficiency. In this paper, we apply the stoc…

New approach improves model generalization through distributionally robust learning.

In this paper, we aim at providing an introduction to the gradient descent based optimization algorithms for learning deep neural network models. Deep learning models involving multiple nonlinear projection layers are very challenging to train. Nowadays, most of the deep learning model training still relies on the back…

Proposes a new method to improve deep model security against adversarial deformations.

Study shows gradient variance increases during deep learning training, contrary to common belief.

The natural gradient of ELBO vanishes in unconstrained optimization, simplifying learning.

Paper proposes recycling model updates in federated learning by exploiting low-rank gradient subspaces.

New methods improve black-box adversarial attacks with fewer queries.

Kernel-Gradient Drifting improves generative modeling for non-Euclidean data.

WMPG reduces policy gradient variance using world models.

Gradient descent and its many variants, including mini-batch stochastic gradient descent, form the algorithmic foundation of modern large-scale machine learning. Due to the size and scale of modern data, gradient computations are often distributed across multiple compute nodes. Unfortunately, such distributed implement…

Mini-batch stochastic gradient descent (SGD) and variants thereof approximate the objective function's gradient with a small number of training examples, aka the batch size. Small batch sizes require little computation for each model update but can yield high-variance gradient estimates, which poses some challenges for…

AdEMAMix optimizer improves model performance and convergence speed.

Unified derivation of high-dimensional linear models using stochastic gradient descent.

Gradient boosting enhances existing Mendelian models for genetic disease risk prediction.

New bounds for model generalization under deterministic gradient descent.

LoRA-One uses one-step full gradient to align adapters for efficient large model fine-tuning.

The paper analyzes how adding gradient data affects overparameterized models' performance.

Backpropagation and the chain rule of derivatives have been prominent; however, the total derivative rule has not enjoyed the same amount of attention. In this work we show how the total derivative rule leads to an intuitive visual framework for creating gradient estimators on graphical models. In particular, previous …