How to model Human Actions distribution with Event Sequence Data
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911196004745216 |
|---|---|
| author | Surkov, Egor Osin, Dmitry Burnaev, Evgeny Shvetsov, Egor |
| author_facet | Surkov, Egor Osin, Dmitry Burnaev, Evgeny Shvetsov, Egor |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_05856 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | How to model Human Actions distribution with Event Sequence Data Surkov, Egor Osin, Dmitry Burnaev, Evgeny Shvetsov, Egor Machine Learning 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. |
| title | How to model Human Actions distribution with Event Sequence Data |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2510.05856 |