Multivariate Uncertainty Quantification with Tomographic Quantile Forests
About
Quantifying predictive uncertainty is essential for safe and trustworthy real-world AI deployment. Yet, fully nonparametric estimation of conditional distributions remains challenging for multivariate targets. We propose Tomographic Quantile Forests (TQF), a nonparametric, uncertainty-aware, tree-based regression model for multivariate targets. TQF learns conditional quantiles of directional projections $\mathbf{n}^{\top}\mathbf{y}$ as functions of the input $\mathbf{x}$ and the unit direction $\mathbf{n}$. At inference, it aggregates quantiles across many directions and reconstructs the multivariate conditional distribution by minimizing the sliced Wasserstein distance via an efficient alternating scheme with convex subproblems. Unlike classical directional-quantile approaches that typically produce only convex quantile regions and require training separate models for different directions, TQF covers all directions with a single model without imposing convexity restrictions. We evaluate TQF on synthetic and real-world datasets, and release the source code on GitHub.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Multivariate distribution estimation | Synthetic Data I | ED0.033 | 35 | |
| Bivariate regression | California housing (10-fold CV) | R20.204 | 7 | |
| Distributional Prediction | sliding-disk synthetic (test) | ED0.29 | 2 |