Subspace Levenberg-Marquardt Algorithms Evaluated for Neural Network Training

Researchers have evaluated subspace Levenberg-Marquardt algorithms for training neural networks on regression and classification tasks. These subspace methods, including Krylov subspace LM and hybrid subspace LM, aim to improve the efficiency of second-order algorithms for larger neural networks. The study compares the performance of these subspace LM variants against the traditional LM algorithm and popular first-order methods like SGD and Adam. AI

影响 This research could lead to more efficient training methods for large neural networks, potentially accelerating development and deployment.

排序理由 The cluster contains a research paper detailing new algorithms for training neural networks. [lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.LG 阅读 →

• Adam
• arXiv
• Krylov subspace LM
• Levenberg-Marquardt Algorithm for Mackey-Glass Chaotic Time Series Prediction
• Neural Networks
• stochastic gradient descent
• subspace LM

• 论文
• 基础设施

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →