🧠 Neural Networks - jhcha.oyo

🎲Probability Academic

arxiv.org·

When Both Layers Learn: Training Dynamics of Representing Linear Models via ReLU Networks

🤖Machine Learning Academic

arxiv.org·

Fourier fractal dimension to predict the generalization of deep neural networks

🤖AI Academic

arxiv.org·

Gradient descent at the Edge of Stability: free energy model and kinetic description of the two-layer network

🧠Deep Learning Academic

arxiv.org·

Uniform Stability and Generalization Error of GD and SGD on Fixed-Point Parameters

📈Optimization Academic

arxiv.org·

Attention at the Theoretical Minimum: A Mathematics of Arrays Framework for Memory-Optimal Transformer Kernels

🤖AI Academic

arxiv.org·

PC Layer: Polynomial Weight Preconditioning for Improving LLM Pre-Training

💬LLMs Academic

arxiv.org·

Closed-Form Spectral Regularization for Multi-Task Model Merging

📈Optimization Academic

arxiv.org·

An Ensembled Latent Factor Model via Differential Evolution and Gradient Descent Optimization

🤖Machine Learning Academic

arxiv.org·

DBHN-Net: Dual-Branch Hybrid Neural Network For Low-Complexity Monaural Speech Enhancement

🤖AI Academic

arxiv.org·

Not Just After One: Sleep-Inspired Replay Prevents Catastrophic Forgetting After Sequential Tasks

🤖AI Academic

arxiv.org·

Quantifying Uncertainty In Wide Two-Layer Neural Networks: On The Law Of The Limiting Fluctuation Process

🤖AI Academic

arxiv.org·

Pretraining Recurrent Networks without Recurrence

🤖AI Academic

arxiv.org·

Machine learning via artificial neural networks coupled with density functional theory and experiments for thermodynamic optimization of high-entropy alloys for hydrogen storage at room temperature

🤖AI Academic

arxiv.org·

Pseudospectral Bounds for Transient Amplification in Coupled Gradient Descent

🤖Machine Learning Academic

arxiv.org·

AI from concrete to abstract: demystifying artificial intelligence to the general public

🤖AI Academic

arxiv.org·

Finite Element-Based Material Learning via Automatic Differentiation: Learning constitutive neural network models from full-field deformation data

🤖AI Academic

arxiv.org·

Gradient Descent with Large Step Size Restores Symmetry in Deep Linear Networks with Multi-Pathway

🤖Machine Learning Academic

arxiv.org·

nnAudio 2: Overcoming Dynamic Compilation Barriers and Transform Inconsistencies

🤖AI Academic

arxiv.org·

Second-Order Path Kernel Interpolation Formulas in Machine Learning

Predictive Coding with Bayesian Priors via Proximal Gradients

When Both Layers Learn: Training Dynamics of Representing Linear Models via ReLU Networks

Fourier fractal dimension to predict the generalization of deep neural networks

Gradient descent at the Edge of Stability: free energy model and kinetic description of the two-layer network

Uniform Stability and Generalization Error of GD and SGD on Fixed-Point Parameters

Attention at the Theoretical Minimum: A Mathematics of Arrays Framework for Memory-Optimal Transformer Kernels

PC Layer: Polynomial Weight Preconditioning for Improving LLM Pre-Training

Closed-Form Spectral Regularization for Multi-Task Model Merging

An Ensembled Latent Factor Model via Differential Evolution and Gradient Descent Optimization

DBHN-Net: Dual-Branch Hybrid Neural Network For Low-Complexity Monaural Speech Enhancement

Not Just After One: Sleep-Inspired Replay Prevents Catastrophic Forgetting After Sequential Tasks

Quantifying Uncertainty In Wide Two-Layer Neural Networks: On The Law Of The Limiting Fluctuation Process

Pretraining Recurrent Networks without Recurrence

Machine learning via artificial neural networks coupled with density functional theory and experiments for thermodynamic optimization of high-entropy alloys for hydrogen storage at room temperature

Pseudospectral Bounds for Transient Amplification in Coupled Gradient Descent

AI from concrete to abstract: demystifying artificial intelligence to the general public

Finite Element-Based Material Learning via Automatic Differentiation: Learning constitutive neural network models from full-field deformation data

Gradient Descent with Large Step Size Restores Symmetry in Deep Linear Networks with Multi-Pathway

nnAudio 2: Overcoming Dynamic Compilation Barriers and Transform Inconsistencies