Attribution methods in the Aurora backend¶
This page documents the methods implemented by geoxplain-aurora-adapter, not a fixed method set imposed by GeoXplain. Other backends may produce results from different explanation libraries or algorithms; the viewer displays their method labels through the same result protocol.
The Aurora backend exposes four single-step methods and one rollout dispatcher. The sections below describe their implemented behavior and the parameters that primarily control cost.
Saliency¶
run_saliency() computes the gradient of the scalar target with respect to each requested input field. Requested variables share one forward/backward computation.
Saliency has no method-specific public parameter. It is the simplest starting point for checking the complete computation and visualization path.
Integrated Gradients¶
run_ig() integrates gradients along a path from a spatially smoothed baseline to the input. The implementation uses the midpoint Riemann rule.
n_steps=32controls the number of integration samples.baseline_sigma_deg=2.5controls Gaussian baseline smoothing in latitude degrees.
Increasing n_steps increases compute approximately linearly.
RISE¶
run_rise() repeatedly masks input regions and measures the target response.
n_masks=200is the number of random masks.cells_h=18andcells_w=36define the low-resolution random mask grid before upsampling.seed=42controls mask generation.baseline_sigma_deg=2.5controls the replacement baseline.
RISE runs separately for each requested input variable, so both n_masks and the number of input variables affect cost.
ViT-CX¶
run_vit_cx() extracts Aurora encoder features, clusters spatial tokens, and measures target changes under cluster occlusion.
hook_stage=2selects encoder stage 0, 1, or 2.n_clusters=256fixes the cluster budget and therefore the number of occlusion forwards per input variable.baseline_sigma_deg=2.5controls the replacement baseline.
Like RISE, ViT-CX runs separately per requested input variable.
Rollout¶
run_rollout() repeatedly advances Aurora by six hours and attributes the resulting autoregressive frames. The current implementation accepts method="saliency" and method="ig". Its timeframes argument is required.
timeframes on the ordinary run_* functions and run_rollout() are not interchangeable: the former computes targets at separate timestamps, while rollout feeds each prediction into the next model step.