
Learning Neural Causal Models from Unknown Interventions
Metalearning over a set of distributions can be interpreted as learning...
read it

A Note on the Estimation Method of Intervention Effects based on Statistical Decision Theory
In this paper, we deal with the problem of estimating the intervention e...
read it

A MetaTransfer Objective for Learning to Disentangle Causal Mechanisms
We propose to metalearn causal structures based on how fast a learner a...
read it

Active Invariant Causal Prediction: Experiment Selection through Stability
A fundamental difficulty of causal learning is that causal models can ge...
read it

Distinguishing Cause from Effect Based on Exogeneity
Recent developments in structural equation modeling have produced severa...
read it

NoisyOR Models with Latent Confounding
Given a set of experiments in which varying subsets of observed variable...
read it

When Causal Intervention Meets Adversarial Examples and Image Masking for Deep Neural Networks
Discovering and exploiting the causality in deep neural networks (DNNs) ...
read it
An Analysis of the Adaptation Speed of Causal Models
We consider the problem of discovering the causal process that generated a collection of datasets. We assume that all these datasets were generated by unknown sparse interventions on a structural causal model (SCM) G, that we want to identify. Recently, Bengio et al. (2020) argued that among all SCMs, G is the fastest to adapt from one dataset to another, and proposed a metalearning criterion to identify the causal direction in a twovariable SCM. While the experiments were promising, the theoretical justification was incomplete. Our contribution is a theoretical investigation of the adaptation speed of simple twovariable SCMs. We use convergence rates from stochastic optimization to justify that a relevant proxy for adaptation speed is distance in parameter space after intervention. Using this proxy, we show that the SCM with the correct causal direction is advantaged for categorical and normal causeeffect datasets when the intervention is on the cause variable. When the intervention is on the effect variable, we provide a more nuanced picture which highlights that the fastesttoadapt heuristic is not always valid. Code to reproduce experiments is available at https://github.com/remilepriol/causaladaptationspeed
READ FULL TEXT
Comments
There are no comments yet.