研究人员开发了一种新的有限内存 BFGS (L-BFGS) 优化算法变体，称为双边 L-BFGS。该方法解决了逆海森近似中条件数爆炸的问题，这可能导致复杂优化景观中的数值不稳定性。双边 L-BFGS 算法使用几何包络动态约束条件数，在保持计算效率和全局收敛特性的同时确保稳定性。 AI
影响 提高了大规模优化问题的数值稳定性和收敛性，可能有利于 AI 模型训练。
排序理由 该集群包含一篇详细介绍新算法及其理论特性的学术论文。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
arXiv:2607.05836v1 Announce Type: cross Abstract: The limited-memory BFGS (L-BFGS) algorithm is a cornerstone of large-scale optimization due to its linear memory and computational costs. However, in ill-conditioned or non-convex landscapes, the implicit inverse Hessian approxima…
The limited-memory BFGS (L-BFGS) algorithm is a cornerstone of large-scale optimization due to its linear memory and computational costs. However, in ill-conditioned or non-convex landscapes, the implicit inverse Hessian approximation can suffer from an exploding condition number…
The limited-memory BFGS (L-BFGS) algorithm is a cornerstone of large-scale optimization due to its linear memory and computational costs. However, in ill-conditioned or non-convex landscapes, the implicit inverse Hessian approximation can suffer from an exploding condition number…