
The LLC algorithm, designed for learning causal relationships in linear cyclic systems with hidden confounders, is inherently non-robust to data contamination, but its robustness can be improved by incorporating robust covariance estimators like MCD and GDE.


coremsg

robustness-analysis-and-improvement-of-the-llc-algorithm-for-causal-discovery-in-linear-cyclic-systems-with-hidden-confounders


Robustness Analysis and Improvement of the LLC Algorithm for Causal Discovery in Linear Cyclic Systems with Hidden Confounders
