Lecture by Eleni Sgouritsa: “Identifying confounders and telling cause…

Eleni Sgouritsa presents her work on Tuesday, November 4th , 12:00pm, Seminar Room I, Institute of Computer Science, Foundation for Research and Technology Hellas (FORTH).

Title:
Identifying confounders and telling cause from effect using latent variable models

Abstract:

Drawing causal conclusions from observed statistical dependencies is a fundamental problem. Conditional independence-based causal discovery (e.g., PC or FCI) cannot be used in case there are no observed conditional independences. Alternative methods investigate a different set of assumptions, namely restricting the function class, e.g., additive noise models. In this talk, I will present two causal inference methods employing different kind of assumptions than the above.

The first is a method to infer the existence and identify a finite confounder of a set of observed dependent variables. It is based on a kernel method to identify finite mixtures of nonparametric product distributions. In the second part, I will focus on the problem of causal inference in the two-variable case (assuming no confounders). A known assumption is used, namely that if X causes Y, P(X) contains no information about P(Y|X). In contrast, P(Y) may contain information about P(X|Y). This asymmetry has implications for common machine learning tasks such as semi-supervised and unsupervised learning, which are employed by the proposed causal inference method.

Short Bio:

Eleni Sgouritsa is in the last year of her PhD at the Max Planck Institute for Intelligent Systems in Tübingen, Germany, under the supervision of Dominik Janzing and Bernhard Schölkopf. She is a member of the causality group which aims at developing novel methods for inferring causal relations from data. Prior to this, she received an MSc and a BSc degree from the Univ. of Houston and the Univ. of Athens, respectively.

Μάθημα R για αρχάριους

Μέλη του Ινστιτούτο Μοριακής Βιολογίας και Βιοτεχνολογίας και του Εργαστηρίου Βιο-Πληροφορικής παρέχουν ένα εισαγωγικό μάθημα στην R, «ανοικτό» για ερευνητές/μεταδιδάκτορες/φοιτητές/προσωπικό του Ιδρύματος Τεχνολογίας και Έρευνας (ΙΤΕ), που ενδιαφέρονται να προσεγγίσουν το λογισμικό και τη γλώσσα προγραμματισμού R, για πρώτη φορά. Τα μαθήματα θα λαμβάνουν χώρα στην αίθουσα Γ100, κάθε Τετάρτη , ώρα 9:15 π.μ.

Talk at MPI for Intelligent Systems in Tuebingen

Pr. Ioannis Tsamardinos is invited to give a talk at MPI for Intelligent Systems in Tuebingen.

Title: Advances in Integrative Causal Analysis

Abstract: We’ll present the concept and approach of Integrative Causal Analysis(INCA), where causal models and relations are induced from multiple,heterogeneous datasets. The datasets maybe heterogeneous in terms of the variable sets they measure or the experimental conditions (interventions)under which they were collected. To solve the problem, all causal models consistent with all datasets are induced and their invariant characteristics (causal features) are found. We introduce a technique where the problem is converted to a SAT (satisfiability) problem that exploits decades of research in satisfiability engines for improved efficiency. In addition, it is also shown that INCA algorithms can in principle also be extended to provide quantitative predictions. The general idea is tested on a couple of dozen real datasets where it is shown that the correlation between two variables never jointly measured, can be predicted based on causal models induced by 2 datasets with overlapping variables. The problem of resolving conflicts arising during learning is also dealt with as well as the problem of identifying the parts of the models that are of high confidence. Finally, new projects and directions are discussed and particular the application of these techniques to learning biological signal pathways from mass cytometry data.