Learning Link-Probabilities in Causal Trees

TL;DR

This paper introduces an incremental and efficient learning algorithm that estimates link probabilities in causal trees using only leaf measurements, even with hidden internal nodes and measurement imprecisions.

Contribution

It presents a novel local and robust method for learning link probabilities in causal trees based solely on leaf data.

Findings

The algorithm accurately estimates link probabilities in causal trees.

It remains robust despite measurement imprecisions.

The method is efficient and incremental.

Abstract

A learning algorithm is presented which given the structure of a causal tree, will estimate its link probabilities by sequential measurements on the leaves only. Internal nodes of the tree represent conceptual (hidden) variables inaccessible to observation. The method described is incremental, local, efficient, and remains robust to measurement imprecisions.