Moment: A Family of Open Time-Series Foundation Models (opens in new tab)

While your eyes can naturally spot a pattern over time in time series data, machines can’t. So as we did for imaging, where we pre-trained models to let machines easily identify objects in pictures, now we (they) are doing the same to let machines “see” patterns over time. Then, how these patterns work, this is another story.

To quote the paper:

The models in this family (1) serve as a building block for diverse time series analysis tasks (e.g., forecasting, classification, anomaly detection, and imputation, etc.), (2) are effective out-of-the-box, i.e., with no (or few) particular task-specific exemplars (enabling e.g., zero-shot forecasting, few-shot classification, etc.), and (3) are tunable using in-distribution and task-specific data to improve performance.

https://en.wikipedia.org/wiki/Time_series

https://en.wikipedia.org/wiki/Foundation_model