arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2510.05856cs.LG
收藏

如何用事件序列数据建模人类行为分布

How to model Human Actions distribution with Event Sequence Data

Applied AI(应用人工智能)

Egor Surkov, Dmitry Osin, Evgeny Burnaev, Egor Shvetsov

AI总结:

本文针对人类行为事件序列预测,挑战自回归范式,提出显式分布预测目标,并引入KL度量量化时间漂移,证明其优于隐式基线,且模式坍缩源于分布不平衡,为建模策略选择提供框架。

摘要:

本文研究人类行为序列中事件未来分布的预测问题,该任务在零售、金融、医疗和推荐系统等领域至关重要,在这些领域中,精确的时间顺序往往不如结果集合重要。我们挑战主流的自回归范式,并探究显式建模未来分布或顺序不变的多令牌方法是否优于保持顺序的方法。我们分析了局部顺序不变性,并引入一种基于KL散度的度量来量化时间漂移。我们发现,一个简单的显式分布预测目标始终优于复杂的隐式基线。我们进一步证明,预测类别的模式坍缩主要由分布不平衡驱动。这项工作为选择建模策略提供了一个原则性框架,并为构建更准确、更稳健的预测系统提供了实用指导。

英文摘要:

This paper studies forecasting of the future distribution of events in human action sequences, a task essential in domains like retail, finance, healthcare, and recommendation systems where the precise temporal order is often less critical than the set of outcomes. We challenge the dominant autoregressive paradigm and investigate whether explicitly modeling the future distribution or order-invariant multi-token approaches outperform order-preserving methods. We analyze local order invariance and introduce a KL-based metric to quantify temporal drift. We find that a simple explicit distribution forecasting objective consistently surpasses complex implicit baselines. We further demonstrate that mode collapse of predicted categories is primarily driven by distributional imbalance. This work provides a principled framework for selecting modeling strategies and offers practical guidance for building more accurate and robust forecasting systems.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑