paper
How Width and Data Shape Generalization Scaling Laws in Quadratic Neural Networks
paperactiveprovisional
how-width-and-data-shape-generalization-scaling-laws-in-quadratic-neural-networks-e4ecebd7·1 events·first seen 17h agoAliases: How Width and Data Shape Generalization Scaling Laws in Quadratic Neural Networks
More like this (12)
Scaling Laws for Neural Language ModelsConservation Laws from Data Symmetry in Neural NetworksAI scaling lawsHumanoid-GPT: Scaling Data and Structure for Zero-Shot Motion TrackingSecond-Order Path Kernel Interpolation Formulas in Machine LearningScaling Laws for Reward Model OveroptimizationGraph Neural Network Encoderpower-law scalingAnatomy of Post-Training: Using Interpretability to Characterize Data and Shape the Learning Signalquantization-aware trainingShannon Scaling LawSubquadratic
Recent events (1)
Theoretical analysis of generalization scaling laws in quadratic two-layer neural networks
A new arXiv preprint derives explicit characterizations of generalization error as a joint function of model width, sample count, and regularization in a quadratic two-layer network with structured data. The analysis reveals a phase diagram with distinct scaling regimes governed by data-dependent power laws tied to the spectral structure of the target function. The work extends scaling law theory beyond fixed-feature or infinite-width regimes by operating in a finite-sample, feature-learning setting, and characterizes interpolation threshold transitions.