Researchers have analyzed the training dynamics of feed-forward neural networks using a Neural Tangent Kernel (NTK) approach applied to Parton Distribution Functions (PDFs) derived from deep inelastic scattering (DIS) data. This method provides an analytical view of how data and methodology influence training, generalizing previous single-flavor analyses to multi-flavor scenarios. The study reveals that gradient-based optimization and parameterization act as regularizers for the ill-posed inverse problem of determining PDFs, with regularization occurring through the incremental inclusion of NTK modes. Notably, only a few leading NTK modes are required to accurately reconstruct the target PDFs, while noise is captured by sub-leading modes learned later in the training process. AI
IMPACT Provides a deeper understanding of neural network training dynamics, potentially applicable to other physics applications and machine learning tasks.
RANK_REASON The item is a research paper detailing a theoretical analysis of neural network training dynamics. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Daily Papers →
- deep inelastic scattering
- feed-forward neural network
- gradient-based optimisation
- Neural tangent kernel
- Parton Distribution Functions
- Shapley value
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
