BriefGPT.xyz
May, 2017
因果生成模型的群不变原理
Group invariance principles for causal generative models
HTML
PDF
Michel Besserve, Naji Shajarisales, Bernhard Schölkopf, Dominik Janzing
TL;DR
本文提出了一个基于群论的框架来研究因果发现算法，统一概括了多种方法中独立原理（ICM）的解释和应用，通过随机泛群转换来评估原因-机制关系并证明该群论视角为研究数据生成机制提供了一个非常通用的工具，并可以直接应用于机器学习。
Abstract
The postulate of
independence of cause and mechanism
(ICM) has recently led to several new
causal discovery
algorithms. The interpretation of independence and the way it is utilized, however, varies across these
→