Latest Articles for Neural Networks

Exploring the role of spiking neural networks in vertical federated learning for data privacy.

A study on improving CNNs using partial labels and adversarial attack models.

This article explores new neural network models for understanding human decision-making in cybersecurity.

Nerva introduces a unique way to build efficient neural networks using sparsity.

A new approach improves GNN reasoning capabilities for complex relationship tasks.

This article discusses improving neural network training speed using transfer learning from supernets.

New neural network architectures improve stability and accuracy in solving partial differential equations.

Examining how deep belief networks can learn from data and create complex representations.

Enhancing teacher models improves student performance in machine learning.

A method to enhance understanding of neural network decision-making.

Examining how inhibitory neurons shape brain function and connectivity.

A look into output range estimation using Simulated Annealing in neural networks.

A comparison of KANs and MLPs in classification performance and hardware efficiency.

New methods improve efficiency in evaluating nonlinear recurrent neural networks for various applications.

A new method enhances neural network performance through better initialization focused on emergence.

Exploring how time structure enhances spiking neural network performance.

This study examines the effectiveness of clean-label physical backdoor attacks in deep neural networks.

New pruning techniques enhance deep learning models for smartphones with limited resources.

CAReLU enhances learning by balancing positive and negative values in deep learning models.

New method improves accuracy and efficiency in deep learning models.

A comparison of SNNs and ANNs for optical flow using event-based camera data.

xAI-Drop leverages explainability to improve Graph Neural Networks' performance and clarity.

BackdoorBench offers a unified approach to assess backdoor learning methods in deep neural networks.

A method to locate and understand weaknesses in language models for improved reliability.

Exploring the capabilities of sinusoidal networks in handling complex data.

Neural networks enhance two-sample testing for complex data analysis.

A new approach to enhance efficiency in multi-task learning using low-rank matrices.

SpikeYOLO enhances SNNs for improved object detection efficiency and accuracy.

Averaging enhances KANs' performance and stability in machine learning tasks.

This research studies the impact of pruning and random structures on RNN performance.

RUM neural networks leverage random walks and memory for enhanced graph processing.

A new framework to assess sparse autoencoders through chess and Othello.

Discover how mood and visuals can enhance AI predictions.

A new method improves emotion prediction in speech analysis.

Exploring improved learning rates in neural networks for scientific computing.

A new benchmark evaluates methods for measuring representational similarity in machine learning.

A new method enhances neural networks' ability to learn and apply knowledge.

This article examines advances in AI beyond transformers.

New methods for better control of RNNs enhance audio effect simulations.

Exploring how diffusion models transform and generate data through systematic noising and de-noising.