Seonglae ChoTime Series Models
Time Series Model Methods
Transformer Model forecasting
- Segmentation of time series into subseries-level patches which are served as input tokens to Transformer
- Channel-independence where each channel contains a single univariate time series that shares the same embedding and Transformer weights across all the series
arxiv.org
https://arxiv.org/pdf/2211.14730
LLM outperforms with structured context
arxiv.org
https://arxiv.org/pdf/2410.18959