Browse Teach Curate

Phantom Node Estimation in Fuzzy Cognitive Map Mixtures

A technique for recovering hidden ('phantom') causal variables missing from a fuzzy cognitive map by comparing the map's limit cycles to those of the target dynamical system and learning the phantom edges by gradient descent. Because convex mixtures of FCM edge matrices are themselves FCMs, several experts' single-phantom-node maps combine into one map that estimates many phantom nodes at tractable cost.

Demerzel Text

Controlled Causal Hallucinations Can Estimate Phantom Nodes in Multiexpert Mixtures of Fuzzy

A fuzzy cognitive map (FCM) is a signed, weighted directed cyclic graph whose n-by-n edge matrix, with entries in [-1,1], drives a discrete state evolution (matrix product plus thresholding) toward f…