QData / spacetimeformer

Multivariate Time Series Forecasting with efficient Transformers. Code for the paper "Long-Range Transformers for Dynamic Spatiotemporal Forecasting."
https://arxiv.org/abs/2109.12218
MIT License
808 stars 191 forks source link

Where you given the location information of NY - TX in the asos model, like lat long #25

Closed jeevanu closed 2 years ago

jeevanu commented 2 years ago

Where you given the location information of NY - TX in the asos model, like lat long

jakegrigsby commented 2 years ago

No, the location information (like lat/long or distance between the stations) isn't given to the model. The model learns the relationships between each variable during training using "variable embeddings" that are randomly initialized. Those embeddings let the attention mechanism distinguish between each variable and learn patterns between them. In the NY-TX case it usually learns the location relationship automatically, and we try to show an example of that in Figure 4 of the paper.

jeevanu commented 2 years ago

Thanks

On Thu, 20 Jan 2022 at 05:39, Jake Grigsby @.***> wrote:

No, the location information (like lat/long or distance between the stations) isn't given to the model. The model learns the relationships between each variable during training using "variable embeddings" that are randomly initialized. Those embeddings let the attention mechanism distinguish between each variable and learn patterns between them. In the NY-TX case it usually learns the location relationship automatically, and we try to show an example of that in Figure 4 of the paper.

— Reply to this email directly, view it on GitHub https://github.com/QData/spacetimeformer/issues/25#issuecomment-1016984971, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADSYZRQCMXC47F3OEJNYZG3UW5HB5ANCNFSM5MH3WT7A . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you authored the thread.Message ID: @.***>