2 problems

0 votes0 replies0 views
The proportional coverage-loss conjecture for split conformal prediction under moving-average dependence
Let $\{W_j\}_{j\in\mathbb{Z}}$ be standard Gaussian variables and let $\epsilon_i=\sum_{j=i-t}^{i}W_j$ define a moving-average process of order $t$ . Suppose $Y_i=f(X_i)+\epsilon_i$ …
Conjectureopenconformal-prediction temporal-dependence time-seriesarXiv posted 2 days ago
0 votes0 replies0 views
Necessity of the stronger signal-to-noise ratio condition for inference
Necessity conjecture. The stronger condition above is necessary for the inference tasks, specifically for establishing uniform tightness of the refined change point estimators in t…
Conjectureopenchange-point-detection high-dimensional-statistics temporal-dependencearXiv posted 2 days ago

The proportional coverage-loss conjecture for split conformal prediction under moving-average dependence