[2005.09136] An Analysis of the Adaptation Speed of Causal Modelsopen searchopen navigation menucontact arXivarXiv Twitter

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 meta-learning criterion to identify the causal direction in a two-variable SCM. While the experiments were promising, the theoretical justification was incomplete. Our contribution is a theoretical investigation of the adaptation speed of simple two-variable 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 cause-effect datasets when the intervention is on the cause variable. When the interv

1 mentions: @LPRmi
Date: 2020/05/21 18:52

Referring Tweets

@LPRmi We just released an analysis of the adaptation speed of causal models t.co/8c1LQOEmFE w/ coauthors Reza Babanezhad, @SimonLacosteJ and Yoshua Bengio.

Related Entries

Read more ignite.contrib.handlers — ignite master documentation
0 users, 0 mentions 2020/01/17 03:51
Read more [1911.03618] Worst Cases Policy Gradientscontact arXivarXiv Twitter
0 users, 1 mentions 2020/02/19 11:20
Read more GitHub - facebookresearch/RTFM: Experiment code for the ICLR 2020 paper "RTFM: Generalising to New E...
0 users, 0 mentions 2020/03/07 15:51
Read more GitHub - cometscome/TightBinding.jl: This can construct the tight-binding model and calculate energi...
0 users, 1 mentions 2020/04/17 03:51
Read more Tenstorrent Unveils Industry's First Dynamic Architecture For Scalable Deep Learning
0 users, 0 mentions 2020/04/20 08:20