Image de couverture de livre

Spectral Edge Dynamics: An Analytical-Empirical Study of Phase Transitions in Neural Network Training

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteur principal:	Xu, Yongzhong
Format:	Preprint
Publié:	2026
Sujets:	Machine Learning Artificial Intelligence
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Spectral Edge Dynamics of Training Trajectories: Signal--Noise Geometry Across Scales
par: Xu, Yongzhong
Publié: (2026)

Feature Repulsion and Spectral Lock-in: An Empirical Study of Two-Layer Network Grokking
par: Xu, Yongzhong
Publié: (2026)

Spectral Edge Dynamics Reveal Functional Modes of Learning
par: Xu, Yongzhong
Publié: (2026)

Optimizer-Induced Low-Dimensional Drift and Transverse Dynamics in Transformer Training
par: Xu, Yongzhong
Publié: (2026)

Spectral Probe-Circuits: A Three-Step Recipe for Identifying Attention-Head Circuits in Pretrained Transformers
par: Xu, Yongzhong
Publié: (2026)

The Geometry of Multi-Task Grokking: Transverse Instability, Superposition, and Weight Decay Phase Structure
par: Xu, Yongzhong
Publié: (2026)

Low-Dimensional and Transversely Curved Optimization Dynamics in Grokking
par: Xu, Yongzhong
Publié: (2026)

Dynamic Spectral Backpropagation for Efficient Neural Network Training
par: Muthuraman, Mannmohan
Publié: (2025)

Global Low-Rank, Local Full-Rank: The Holographic Encoding of Learned Algorithms
par: Xu, Yongzhong
Publié: (2026)

Early-Warning Signals of Grokking via Loss-Landscape Geometry
par: Xu, Yongzhong
Publié: (2026)

Gradient-Direction Sensitivity Reveals Linear-Centroid Coupling Hidden by Optimizer Trajectories
par: Xu, Yongzhong
Publié: (2026)

When Do Attention Circuits Form? Developmental Trajectories of Capability and Attention-Sink Emergence Across Three 1B-ClassArchitectures
par: Xu, Yongzhong
Publié: (2026)

Low-Dimensional Execution Manifolds in Transformer Learning Dynamics: Evidence from Modular Arithmetic Tasks
par: Xu, Yongzhong
Publié: (2026)

Spectral Theory for Edge Pruning in Asynchronous Recurrent Graph Neural Networks
par: Bessone, Nicolas
Publié: (2025)

The Lifecycle of the Spectral Edge: From Gradient Learning to Weight-Decay Compression
par: Xu, Yongzhong
Publié: (2026)

Catapult Dynamics and Phase Transitions in Quadratic Nets
par: Meltzer, David, et autres
Publié: (2023)

Spectral Higher-Order Neural Networks
par: Peri, Gianluca, et autres
Publié: (2026)

Spatio-Spectral Graph Neural Networks
par: Geisler, Simon, et autres
Publié: (2024)

Addressing Spectral Bias of Deep Neural Networks by Multi-Grade Deep Learning
par: Fang, Ronglong, et autres
Publié: (2024)

An Analytic Solution to Covariance Propagation in Neural Networks
par: Wright, Oren, et autres
Publié: (2024)

Physics-inspired Energy Transition Neural Network for Sequence Learning
par: Wu, Zhou, et autres
Publié: (2025)

Grokking as a Variance-Limited Phase Transition: Spectral Gating and the Epsilon-Stability Threshold
par: Acharya, Pratyush, et autres
Publié: (2026)

From Density Matrices to Phase Transitions in Deep Learning: Spectral Early Warnings and Interpretability
par: Hennick, Max, et autres
Publié: (2026)

Edge Prompt Tuning for Graph Neural Networks
par: Fu, Xingbo, et autres
Publié: (2025)

Dynamic DropConnect: Enhancing Neural Network Robustness through Adaptive Edge Dropping Strategies
par: Yang, Yuan-Chih, et autres
Publié: (2025)

Spectral Greedy Coresets for Graph Neural Networks
par: Ding, Mucong, et autres
Publié: (2024)

Piecewise Constant Spectral Graph Neural Network
par: Martirosyan, Vahan, et autres
Publié: (2025)

Spectral Entropy Collapse as a Phase Transition in Delayed Generalisation: An Interventional and Predictive Framework for Grokkin
par: Khanh, Truong Xuan, et autres
Publié: (2026)

NeuZip: Memory-Efficient Training and Inference with Dynamic Compression of Neural Networks
par: Hao, Yongchang, et autres
Publié: (2024)

ENADPool: The Edge-Node Attention-based Differentiable Pooling for Graph Neural Networks
par: Zhao, Zhehan, et autres
Publié: (2024)

SmartMixed: A Two-Phase Training Strategy for Adaptive Activation Function Learning in Neural Networks
par: Omidvar, Amin
Publié: (2025)

Empirical Guidelines for Deploying LLMs onto Resource-constrained Edge Devices
par: Qin, Ruiyang, et autres
Publié: (2024)

GraphEdge: Dynamic Graph Partition and Task Scheduling for GNNs Computing in Edge Network
par: Xiao, Wenjing, et autres
Publié: (2025)

Rethinking Spectral Graph Neural Networks with Spatially Adaptive Filtering
par: Guo, Jingwei, et autres
Publié: (2024)

Advancing On-Device Neural Network Training with TinyPropv2: Dynamic, Sparse, and Efficient Backpropagation
par: Rüb, Marcus, et autres
Publié: (2024)

IDInit: A Universal and Stable Initialization Method for Neural Network Training
par: Pan, Yu, et autres
Publié: (2025)

Native Fortran Implementation of TensorFlow-Trained Deep and Bayesian Neural Networks
par: Furlong, Aidan, et autres
Publié: (2025)

A Convexity-dependent Two-Phase Training Algorithm for Deep Neural Networks
par: Hrycej, Tomas, et autres
Publié: (2025)

Refined Edge Usage of Graph Neural Networks for Edge Prediction
par: Jin, Jiarui, et autres
Publié: (2022)

Graph Neural Network with One-side Edge Sampling for Fraud Detection
par: Trieu, Hoang Hiep
Publié: (2026)