For linear interventional SDEs, r interventions identify the drift almost surely while r-2 cannot; in the small-noise nonlinear case, r+1 interventions identify the low-rank factors up to permutation and scaling.
With this setup, we obtain a family of distributions{ρ k}K k=0 in gene expression space
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Towards Identifiability of Interventional Stochastic Differential Equations
For linear interventional SDEs, r interventions identify the drift almost surely while r-2 cannot; in the small-noise nonlinear case, r+1 interventions identify the low-rank factors up to permutation and scaling.