本论文介绍了三种提高神经控制微分方程（NCDE）训练效率和可扩展性的方法，NCDE是一类用于连续时间序列数据的模型。提出的技术包括用于更快逼近的Log-NCDE、用于闭式解和并行计算的Linear NCDE，以及用于进一步提高效率的Structured Linear NCDE。这些进展共同将训练时间缩短了多达三个数量级，同时在时间序列基准测试中取得了最先进的结果。 AI
影响 这些进展可能能够更有效地训练连续时间模型，从而提高复杂时间序列任务的性能。
排序理由 该集群包含一篇学术论文，详细介绍了改进机器学习模型训练的新方法。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
arXiv:2607.05280v1 Announce Type: new Abstract: Many real-world systems evolve continuously, yet most machine learning models interpret time series as discrete sequences. Continuous-time approaches instead treat time series as samples from an underlying input path, a formulation …
Many real-world systems evolve continuously, yet most machine learning models interpret time series as discrete sequences. Continuous-time approaches instead treat time series as samples from an underlying input path, a formulation that naturally accommodates irregularly sampled …