Understanding the initial condensation of convolutional neural networks

DOI10.4208/CSIAM-AM.SO-2024-0011MaRDI QIDQ6948233zbMATH OpenFDO

Authors Zhangchen Zhou, Hanxu Zhou, Yuqing Li, Zhi-Qin John Xu

Publication date 5 August 2025

Published in CSIAM Transactions on Applied Mathematics (Search for Journal in Brave)

Full work available at URL https://doi.org/10.4208/CSIAM-AM.SO-2024-0011

zbMATH Keywords

condensationconvolutional neural networkdynamical regime

Mathematics Subject Classification ID

Theoretical approximation of solutions to ordinary differential equations (34A45) Computing methodologies and applications (68U99) Nonconvex programming, global optimization (90C26)

Cites work

• 10.1162/153244303321897690
• A comparative analysis of optimization and generalization properties of two-layer neural network and random feature models under gradient descent dynamics
• Adaptive estimation of a quadratic functional by model selection.
• Frequency Principle: Fourier Analysis Sheds Light on Deep Neural Networks
• High-dimensional dynamics of generalization error in neural networks
• Mean field analysis of neural networks: a central limit theorem
• Phase diagram for two-layer ReLU neural networks at infinite-width limit
• Understanding the initial condensation of convolutional neural networks
• Universality of deep convolutional neural networks

Cited in

(2)

• Understanding the initial condensation of convolutional neural networks
• Anchor function: a type of benchmark functions for studying language models

This page was built for publication: Understanding the initial condensation of convolutional neural networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6948233)