Helmholtz Gemeinschaft

Search
Browse
Statistics
Feeds

Using unlabeled data to discover bivariate causality with deep restricted Boltzmann machines

Item Type:Article
Title:Using unlabeled data to discover bivariate causality with deep restricted Boltzmann machines
Creators Name:Sokolovska, N., Permiakova, O., Forslund, K. and Zucker, J.D.
Abstract:An important question in microbiology is whether treatment causes changes in gut flora, and whether it also affects metabolism. The reconstruction of causal relations purely from non-temporal observational data is challenging. We address the problem of causal inference in a bivariate case, where the joint distribution of two variables is observed. In this contribution, we introduce a novel method of causality discovering which is based on the widely used assumption that if X causes Y, then P(X) and P(Y|X) are independent. We propose to explore a semi-supervised approach where P(Y|X) and P(X) are estimated from labeled and unlabeled data respectively, whereas the marginal probability is estimated potentially from much more unlabeled data than the conditional distribution. We illustrate by experiments on several benchmarks of biological network reconstruction that the proposed approach is very competitive in terms of computational time and accuracy compared to the state-of-the-art methods. Finally, we apply the proposed method to an original medical task where we study whether drugs confound human metagenome.
Keywords:Causal Inference, Semi-Supervised Learning, Probabilistic Models, Metagenomic Data
Source:IEEE/ACM Transactions on Computational Biology and Bioinformatics
ISSN:1545-5963
Publisher:IEEE Computer Society
Volume:17
Number:1
Page Range:358-364
Date:January 2020
Official Publication:https://doi.org/10.1109/TCBB.2018.2879504
PubMed:View item in PubMed

Repository Staff Only: item control page

Open Access
MDC Library