Latest Articles for Neural Networks

Research reveals complexities in deep neural networks beyond traditional models.

This paper analyzes multi-index models and their role in learning from data.

New benchmark tool assesses discrete audio tokens for various speech processing tasks.

This study reveals how language models change behavior during training.

Examining how transformer models improve with size and complexity.

Study analyzes generalization and performance of random feature ridge regression using eigenvalues.

A study on improving neural network training with non-differentiable activation functions.

Introducing SeTAR, a training-free solution for detecting out-of-distribution data in neural networks.

Exploring the benefits of repeated data in training neural networks.

This article discusses how deep neural networks learn language through next-token prediction.

Examining how prompts affect reasoning in large language models.

This study examines how equivariant neural networks enhance Offline RL performance using limited data.

This article discusses how neuron models help analyze complex brain activity.

QuEE combines quantization and early exiting for efficient machine learning.

A new approach improves optimization of complex loss functions in neural networks.

A new method predicts probe positions for clearer imaging in ptychography.

A look into how linear networks learn and evolve during training.

Combining physics and geometry for improved acoustic scattering predictions.

Discover how Leaky ResNets enhance deep learning techniques.

A look into injectivity challenges and methods in ReLU layers within neural networks.

Introducing DARE, a method to improve machine learning without forgetting old knowledge.

A novel approach enhances Transformer models for better long text processing.

A look at the role of complexity in model performance.

A novel loss function enhances feature learning in classification tasks.

Exploring classification methods for overlapping Gaussian mixtures in machine learning.

A groundbreaking model handles dynamic graphs while boosting performance and reducing training time.

Examining how normalization layers influence transformer performance and task handling.

This study uses sparse autoencoders to interpret attention layer outputs in transformers.

New methods improve modeling of electromagnetic problems with interfaces using neural networks.

A new neural network approach improves accuracy in hyperbolic conservation laws.

How Mixtures of Experts enhance performance in Deep Reinforcement Learning tasks.

Using neural networks on FPGAs to enhance high-speed communication reliability.

Exploring the role of neurons in enhancing IR model interpretability.

Introducing a new approach to enhance video data representation and efficiency.

Examining the impact of attention masks and layer normalization on transformer models.

PointTree offers an innovative solution for accurately reconstructing neuron connections in the brain.

Exploring the latest developments in models for processing long sequences of data.

This study examines how task similarity affects continual learning in neural networks.

This study examines how model size affects performance in Online Continual Learning.

This research explores how to manage neural activities with advanced control methods.