Residual Prior
A model that learns the residual, defined as the true motion minus a baseline prediction such as harmonic extension, on top of that baseline. It compensates for the baseline method's inability to capture high-frequency detail.
Timeline (last 90 days)
Residual amplitudes can differ by more than an order of magnitude across different geometric families and collapse sharply as observations become denser, resulting in a Scaling Barrier for cross-family training.
A residual prior refers to learning a network on top of a baseline prediction to predict the difference between the true motion and the baseline, and its statistical properties depend heavily on the geometric family and observation density.
All Facts (2)
Residual amplitudes can differ by more than an order of magnitude across different geometric families and collapse sharply as observations become denser, resulting in a Scaling Barrier for cross-family training.
50%UnverifiedA residual prior refers to learning a network on top of a baseline prediction to predict the difference between the true motion and the baseline, and its statistical properties depend heavily on the geometric family and observation density.
50%