Latest Articles for Neural Networks

Virtual nodes help improve performance in graph neural networks by enhancing information flow.

A new method to improve learning retention in AI systems.

Introducing AnyLoss, transforming metrics into loss functions for better model training.

New method improves data removal in dynamic graph models while ensuring privacy.

A new method ensures reliable image restoration by training monotonic neural networks.

MoEUT improves Universal Transformers' efficiency and performance in language tasks.

MLPs show surprising effectiveness in in-context learning, challenging views on model complexity.

SMT optimizes fine-tuning of large language models with reduced resource demands.

A method to train large neural networks efficiently while using less memory.

New methods aim to improve machine learning by retaining knowledge while adapting to new tasks.

ETNNs enhance complex data analysis through topological and geometric integration.

A new hybrid system combines optical and electronic methods for efficient image classification.

This article discusses TULIP, a method for better uncertainty estimation in machine learning.

New methods improve stability of control systems under uncertain conditions.

Learn how Transfer Entropy enhances Convolutional Neural Networks training and performance.

Research reveals how large language models respond to various input types.

New method reduces backdoor threats in deep neural networks.

This article examines U-Nets and their role in image processing using generative models.

Explore the impact of norms on neural networks' training and performance.

An analysis of factors influencing forgetting in machine learning.

This study explores how neural network representations evolve during training, inspired by nature.

This study explores how DNNs learn and adapt through training.

This article discusses how GRSNN enhances graph reasoning tasks using synaptic delay.

Learn how hyperparameters impact training in wide neural networks.

An analysis of SGD behavior in machine learning with insights on eigenvalues and training stability.

Exploring new methods for designing frames in machine learning.

A look into neural collapse and its impact on deep learning models.

Exploring the benefits and applications of EQCNNs in machine learning.

Examining the effects of outlier features on neural network training.

Exploring how Riemannian Geometry reshapes our understanding of neural networks.

This research investigates the role of latent variables in Transformers' performance.

This article discusses challenges in few-shot fine-tuning of diffusion models and solutions.

A look at the roles of injectivity and surjectivity in ReLU networks.

A new approach to offline reinforcement learning improves policy learning using diffusion models.

A new approach for generating programs based on images using advanced neural models.

A new approach to improve efficiency in neural architecture search processes.

Research on optimizing deep learning models with sparsity and quantization techniques.

This study investigates how small changes can mislead CNNs in critical tasks.

Exploring advanced methods for effective graph data analysis.

New model improves long-range information flow in graph data.