Table 2
Change in the Leave-One-Out Cross-Validation and Widely Applicable Information Criterion statistics between models compared with the best-fitting model, which is the piecewise-Lambertian model of Hu et al. (2015).
| LOO | WAIC | |||
|---|---|---|---|---|
| Model | Δ | Weight | Δ | Weight |
| Piecewise-Lam | – | 0.62 | – | 0.62 |
| Sinusoid | 4 | 0.38 | 4 | 0.38 |
| Lambert | 48 | 0.00 | 48 | 0.00 |
| No var. | 57 | 0.00 | 57 | 0.00 |
Notes. The smaller the ΔLOO or ΔWAIC, the more compatible the data are with a given model. The weight for each statistic can be interpreted as the probability that a given model is preferred over the others. The statistic values and probabilities evaluated with both techniques are quite similar. The Piecewise-Lambertian model is similarly compatible with the sinusoidal model, but the sinusoidal model lacks a physical interpretation so we favor the Piecewise-Lambertian. The pure Lambertian and no-variation models are ruled out.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.