Generalizable autoregressive modeling of time series through functional narratives
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929537640562688 |
|---|---|
| author | Liu, Ran Ma, Wenrui Zippi, Ellen Pouransari, Hadi Xiao, Jingyun Sandino, Chris Mahasseni, Behrooz Minxha, Juri Azemi, Erdrin Dyer, Eva L. Moin, Ali |
| author_facet | Liu, Ran Ma, Wenrui Zippi, Ellen Pouransari, Hadi Xiao, Jingyun Sandino, Chris Mahasseni, Behrooz Minxha, Juri Azemi, Erdrin Dyer, Eva L. Moin, Ali |
| contents | Time series data are inherently functions of time, yet current transformers often learn time series by modeling them as mere concatenations of time periods, overlooking their functional properties. In this work, we propose a novel objective for transformers that learn time series by re-interpreting them as temporal functions. We build an alternative sequence of time series by constructing degradation operators of different intensity in the functional space, creating augmented variants of the original sample that are abstracted or simplified to different degrees. Based on the new set of generated sequence, we train an autoregressive transformer that progressively recovers the original sample from the most simplified variant. Analogous to the next word prediction task in languages that learns narratives by connecting different words, our autoregressive transformer aims to learn the Narratives of Time Series (NoTS) by connecting different functions in time. Theoretically, we justify the construction of the alternative sequence through its advantages in approximating functions. When learning time series data with transformers, constructing sequences of temporal functions allows for a broader class of approximable functions (e.g., differentiation) compared to sequences of time periods, leading to a 26\% performance improvement in synthetic feature regression experiments. Experimentally, we validate NoTS in 3 different tasks across 22 real-world datasets, where we show that NoTS significantly outperforms other pre-training methods by up to 6\%. Additionally, combining NoTS on top of existing transformer architectures can consistently boost the performance. Our results demonstrate the potential of NoTS as a general-purpose dynamic learner, offering a viable alternative for developing foundation models for time series analysis. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_08421 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Generalizable autoregressive modeling of time series through functional narratives Liu, Ran Ma, Wenrui Zippi, Ellen Pouransari, Hadi Xiao, Jingyun Sandino, Chris Mahasseni, Behrooz Minxha, Juri Azemi, Erdrin Dyer, Eva L. Moin, Ali Machine Learning Time series data are inherently functions of time, yet current transformers often learn time series by modeling them as mere concatenations of time periods, overlooking their functional properties. In this work, we propose a novel objective for transformers that learn time series by re-interpreting them as temporal functions. We build an alternative sequence of time series by constructing degradation operators of different intensity in the functional space, creating augmented variants of the original sample that are abstracted or simplified to different degrees. Based on the new set of generated sequence, we train an autoregressive transformer that progressively recovers the original sample from the most simplified variant. Analogous to the next word prediction task in languages that learns narratives by connecting different words, our autoregressive transformer aims to learn the Narratives of Time Series (NoTS) by connecting different functions in time. Theoretically, we justify the construction of the alternative sequence through its advantages in approximating functions. When learning time series data with transformers, constructing sequences of temporal functions allows for a broader class of approximable functions (e.g., differentiation) compared to sequences of time periods, leading to a 26\% performance improvement in synthetic feature regression experiments. Experimentally, we validate NoTS in 3 different tasks across 22 real-world datasets, where we show that NoTS significantly outperforms other pre-training methods by up to 6\%. Additionally, combining NoTS on top of existing transformer architectures can consistently boost the performance. Our results demonstrate the potential of NoTS as a general-purpose dynamic learner, offering a viable alternative for developing foundation models for time series analysis. |
| title | Generalizable autoregressive modeling of time series through functional narratives |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2410.08421 |