Get Started
Home
Topics
Search
Library
Research questionHow can federated LiNGAM causal discovery remain reliable under near-symmetric noise without centralizing data?Federated causal discovery must combine information held by separate clients without pooling regulated data. Near-symmetric noise makes existing cumulant-based methods unreliable, while variable scaling can cause them to reflect variance structure rather than the intended causal asymmetry.
AI
Machine Learning
Research Paper
Statistical Machine Learning
Latest papersRecent research connected to this question, newest first.Federated Causal Discovery via Regression-Directed CumulantsThe source concerns federated LiNGAM under horizontal, vertical, and hybrid partitions. It studies higher-order cumulant methods, including single-round procedures and exact unlearning from individual observations or clients, and reports failures under near-symmetric noise, near-random ordering after marginal standardisation, and variance-ladder effects at deployment-sized samples. Scale-invariant DirectLiNGAM is used as a comparison but is not federable under the described protocol.research paper · Sep 3, 2026
Related questions
How can federated learning protect client updates from inference while resisting Byzantine manipulation?How can complex causal queries be identified under non-IID data, transfer, and missing observations?How can federated learning stay reliable when heterogeneous edge clients are intermittently available and availability changes over time?How can differentiable causal discovery override wrong edge priors while preserving directional identifiability?