Latest Articles for Gradients

A novel method that uses TEEs to protect machine learning models in federated learning.

A new algorithm enhances accuracy in solving backward stochastic differential equations.

ModelGiF offers a method to quantify relationships between deep learning models.

Examining energy minimizers and their properties in varied regions.

Variance suppression enhances deep neural network performance in challenging data conditions.

Learn how preconditioning enhances optimization methods for faster solutions.

SAM improves neural network training by focusing on parameter stability.

Improving optimization through limited memory gradient techniques for better performance.

A framework to make neural networks' decisions easier to understand.

A new method to assess model accuracy without true labels.

Introducing a method to optimize functions with inexact gradients and Hessians.

A detailed comparison of GMLS and RBF-FD methods for accurate surface derivative calculations.

ELRA adapts learning rates to improve machine learning efficiency.

Quantum computing shows promise in machine learning through qGANs, despite training challenges.

Discover how Actions World Models improve decision-making in complex environments.

New method optimizes sampling by combining it with optimization techniques.

Exploring the role of Hamming-weight-preserving circuits in enhancing quantum machine learning.

This study examines how language models learn and store information during training.

Explore the role of Poincaré inequalities in numerical methods for partial differential equations.

New methods improve training speed and efficiency for large language models.

Introducing novel techniques for interpreting complex-valued neural networks in various applications.

This article introduces a new method to improve stability in real-time control systems.

Learn how SGD with Clipping enhances gradient estimation in noisy data environments.

A new method enhances learning by addressing varying data scales.

A new method enhances OOD detection by focusing on gradient information.

A new loss function improves accuracy in semantic segmentation tasks.

RLLC enhances traditional optimizers by using memory units for better performance.

Exploring methods to recover model parameters from leverage scores in regression analysis.

Integrating deep learning enhances seismic imaging accuracy and efficiency.

Learn how the Heavy Ball method enhances optimization techniques.

Methods to handle outliers effectively in data analysis.

Learn about methods to find optimal solutions in various fields.

VSGD combines traditional methods with probabilistic modeling for better deep learning optimization.

A novel approach improves distance optimization strategies at singular points.

Research proposes methods for better gradient estimation in quantum computing.

New methods improve efficiency in face morphing using diffusion models.

Research shows vulnerabilities in federated learning's approach to text privacy.

A new approach enhances the effectiveness of attacks on safety-focused language models.

SDPRLayer improves robot learning through effective optimization methods.

A new method tackles wireless errors in federated learning for IoT devices.