Latest Articles for Neural Networks

Exploring how pruning and quaternions enhance neural network performance.

This study reveals SGD's advantages in robustness over adaptive training methods.

This paper discusses Bayesian Flow Networks and their effectiveness in data generation.

A new method enhances how neural networks handle unseen data.

Ternary SVD offers a simpler and efficient way to compress neural networks.

A look into neural transport control and its applications.

Introducing reversible SNN nodes for efficient training and memory use.

Advancements in deep learning enhance Cone Beam CT imaging quality and efficiency.

New techniques improve the training process for spiking neural networks.

A new framework improves deep learning model reliability against subtle changes.

EDNIL enhances model robustness against changes in data contexts.

A novel approach allows training neural networks efficiently on small devices without backward propagation.

HollowNeRF improves efficiency and quality in 3D rendering using targeted data management.

A look into the injectivity of ReLU layers and their implications in neural networks.

Research on reservoir computing focuses on optimizing network designs for improved machine learning.

Evol-Q enhances quantization accuracy in Vision Transformers through evolutionary search techniques.

This article explores improving classification using hierarchical softmax in machine learning.

Examining how Causal Neural Networks enhance resilience to adversarial attacks.

A new metamaterial neural network design improves efficiency and speed in image processing.

Explore initial training strategies in class-incremental learning without retaining past examples.

Exploring efficient models for similarity matching in machine learning.

Equivariance helps neural networks recognize images consistently despite transformations.

Exploring the potential and challenges of optical neural networks using light for computation.

This article discusses the Lottery Ticket Hypothesis and pruning techniques in DNNs.

A look at improving deep RNN training through gradient stability methods.

A groundbreaking approach to training neural networks without gradients.

A fresh method aims to help models learn new tasks without forgetting old ones.

A new method enhances diffusion models for clearer image generation.

A novel method to enhance neural networks by managing neuron correlations.

An overview of Hopfield networks addressing generalization and overfitting.

Researchers develop weight masks to help models retain knowledge while learning new tasks.

A new method enhances multiclass classification performance using ECOC with NEAT.

A new method enhances the classification of underwater sounds from vessels using neural networks.

A new tool estimates how neural networks react to input changes, crucial for safety.

A new method to improve speech quality using energy-efficient networks.

A look into the dynamics of SGD and its effects on model training.

A new method enhances models' performance on unseen data in computer vision.

A new method enhances model performance across various data types.

Examining how parameter isolation improves continual learning through dynamic sparse training methods.

Exploring how adaptive features improve neural network performance during training.