What cost function does CrossWave use for model tuning is it RMSE?

Design and Planning
,
1

What cost function does CrossWave use for model tuning—is it RMSE?

Hi everyone,

I’m trying to understand the cost function used by CrossWave during model tuning in Atoll. Does it directly minimize the RMSE between predicted and measured signal levels, or does it use a different cost function?

We are obtaining the same reported RMSE for the following configurations using the same clutter data:

  • SPM
  • CrossWave Base Model
  • CrossWave with Standard Tuning
  • CrossWave with Advanced Tuning

We expected some variation, particularly between the base model and the tuned configurations, but the reported RMSE remains unchanged.

Could anyone help clarify the following?

  1. What cost function does CrossWave use for Standard and Advanced Tuning? Do these modes optimize the same metric?
  2. Is the RMSE reported in the model evaluation statistics the actual metric minimized during tuning?
  3. What settings or checks would you recommend to investigate why all four configurations produce the same reported RMSE?

Any insights, practical experience, or references to relevant documentation would be greatly appreciated.

Thanks!