Significance of predicting SS and ES nodes?

Good evening,

I am joining the competition a little later than most I assume, but I had a question regarding the significance of the SS (start of the study period) and ES (end of the study period) for the ML approach.

As I understand it, these labels are simply meant to indicate the first and last time indices for each object during the observation period but they have no reflection of the satellite’s behavior. Therefore, any machine learning based approach will be faulty and the performance metrics will not be an accurate representation of the model’s ability to detect the 3 maneuver movement node labels (ID, AD, and IK). Could you provide insight into how these 2 nodes correlate to behavior patterns?

Thank you in advance,

  • The AXOLTL team

Hi,

as indicated here, the SS node simply serves as an entry point node. As it is not guaranteed that a mode change has occured here (and as you correctly said, it would be impossible to know for sure without more historical data), it is simply set to ‘SS’. The ‘ES’ nodes are completely ignored during evaluation, so you can cut them off.

How you handle this in your ML solution is something you will have to figure out yourself - a simple solution would be to manually set the Node to ‘SS’ for every place where the TimeIndex is 0.

Best

1 Like

Thank you for your reply David!