Research map / Probabilistic and scientific modeling / Variational and generative sampling
The Computational Advantage of Depth in Learning High-Dimensional Hierarchical Targets
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Theoretical study of why deep networks trained by gradient descent outperform shallow ones, using single- and multi-index Gaussian hierarchical target functions with a hierarchy of latent subspace dimensionalities.
Problem
Understanding the advantages of deep neural networks trained by gradient descent over shallow models remains an open theoretical challenge.
Approach
Introduces a class of target functions (single and multi-index Gaussian hierarchical targets) with a hierarchy of latent subspace dimensionalities, enabling analytic study of learning dynamics and generalization of deep versus shallow networks trained by GD.
Closest papers by meaning
- How Deep Networks Learn Sparse and Hierarchical Data: the Sparse Random Hierarchy Model (ICML 2024 · similarity 0.87)
- Feature Learning Dynamics in Infinite-Depth Neural Networks (NeurIPS 2026 · similarity 0.86)
- Does SGD really happen in tiny subspaces? (ICLR 2025 · similarity 0.85)
- Investigating the Overlooked Hessian Structure: From CNNs to LLMs (ICML 2025 · similarity 0.85)
- Tensor Programs VI: Feature Learning in Infinite Depth Neural Networks (ICLR 2024 · similarity 0.85)
- Deep Linear Network Training Dynamics from Random Initialization: Data, Width, Depth, and Hyperparameter Transfer (ICML 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.