Causal Inference

"I would rather discover one true cause than gain the kingdom of Persia. "

Democritus

Sign up here for the emailing list.

Weekly meeting schedule

Spring 2019

We will start late in Spring 2019 as I (Jakob) will be travelling, including to the Machine Learning Summer School 2019 in Stellenbosch which has a big focus this year on causality, more so than ever, see here: https://mlssafrica.com/programme-schedule/

(Also, see: https://jakobzeitler.weebly.com/blog/machine-learning-summer-school-mlss-africa-2019)

Talks, Conferences, Workshops

January 7th-18th: Machine Learning Summer School, Stellenbosch, South Africa

February 8th: iSchool: Introduction to Causal Inference, Syracuse, USA

March 3rd-7th: Workshop on Causality (Jonas Peters, Vanessa Didelez), Uder, Germany

May 22nd-24th: Atlantic Causal Inference Conference 2019, Montreal, Canada

Week 1

Time: 10.30AM, TBA, estimated mid February

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

Fall 2018

Week 10

Time: 10.30AM, Monday 3rd of December

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

TBA

Thanksgiving break

No meetings.

Week 9

Time: 10.30AM, Monday 5th of November

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our seventh meeting for the Fall 2018, we will look at the Hilbert-Space Independence Criterion and how to construct a p-value for it, as described in "Large-scale kernel methods for independence testing" by Gretton et al.

Week 8

Time: 10.30AM, Monday 29th of October

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our seventh meeting for the Fall 2018, we will look at the Hilbert-Space Independence Criterion and how to construct a p-value for it, as described in "Large-scale kernel methods for independence testing" by Gretton et al.

Week 7

Time: 10.30AM, Monday 22nd of October

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our sixth meeting for the Fall 2018, we will look at the Hilbert-Space Independence Criterion and how to construct a p-value for it, as described in "A Kernel Statistical Test of Independence" by Gretton et al. (http://www.kyb.mpg.de/fileadmin/user_upload/files/publications/attachments/NIPS2007-Gretton_%5b0%5d.pdf)

Week 6

Time: 10.30AM, Monday 15th of October

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our fifth meeting for the Fall 2018, we will look at "A Kernel Statistical Test of Independence" by Gretton et al. (http://www.kyb.mpg.de/fileadmin/user_upload/files/publications/attachments/NIPS2007-Gretton_%5b0%5d.pdf)

Week 4

Time: 10.30AM, Monday 1st of October

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our foruth meeting for the Fall 2018, we will look at "DeepMatch: Balancing Deep Covariate Representations for Causal Inference Using Adversarial Training" by Nathan Kallus. ( https://arxiv.org/abs/1802.05664)

Week 3

Time: 10.30AM, Monday 24th of September

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our third meeting for the Fall 2018, we will look at the MIT MIMIC dataset as well as some more work in our Jupyter notebooks on ANM and Information Complexity models .

Week 2

Time: 10.30AM, Monday 17th of September

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our second meeting for the Fall 2018, we will look at some recent work in our Jupyter notebooks on ANM and Information Complexity models.

Week 1

Time: 10.30AM, Monday 10th of September

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

Respond for attendance to: jkzeitle@syr.edu

In our first meeting for the Fall 2018, we will have an open discussion on anything related to causality. Bring your thoughts on deep learning, Philosophy of Science, Econometrics and the coming Syracuse winter. Snacks will be provided.

Summer 2018

Week 11

Time: undetermined, contact organizer

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

We managed to reproduce the results of Hoyer et al. 2009 and the Old Faithful example, but encountered different p-values than in the paper. Therefore, we need to analyse the code and figure out where the implementation changed and rewrite a version ourselves.

Here is our reproduction:

Week 10

Time: undetermined, contact organizer

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our tenth meeting we will discuss different attempts (in Matlab, R and Python) of the Old Faithful example as mentioned in "Nonlinear causal discovery with additive noise models" by Hoyer et al. (link: https://papers.nips.cc/paper/3548-nonlinear-causal-discovery-with-additive-noise-models)

Week 9

Time: 1 PM, Tuesday 21st of August

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our ninth meeting we will discuss independence tests and attempt our own solution for the Old Faithful example as mentioned in "Nonlinear causal discovery with additive noise models" by Hoyer et al. (link: https://papers.nips.cc/paper/3548-nonlinear-causal-discovery-with-additive-noise-models)

Week 8

Time: 1 PM, Tuesday 7th of August

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our eighth meeting we will discuss HSICs and walk through the code for the Old Faithful example as mentioned in "Nonlinear causal discovery with additive noise models" by Hoyer et al. (link: https://papers.nips.cc/paper/3548-nonlinear-causal-discovery-with-additive-noise-models)

Week 7

Time: 1 PM, Tuesday 31st of July

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our seventh meeting we will discuss "Nonlinear causal discovery with additive noise models" by Hoyer et al. (link: https://papers.nips.cc/paper/3548-nonlinear-causal-discovery-with-additive-noise-models)

Week 6

Time: 1 PM, Tuesday 24th of July

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our sixth meeting we will discuss the assumptions for Structural Identifiability in Section 4.1.1 to 4.1.4 (page 43-52) of “Elements of Causal Inference” (link: https://www.dropbox.com/s/o4345krw428kyld/11283.pdf?dl=0) by Bernhard Schölkopf, Jonas Peters and Dominik Janzing.

Week 5

Time: 1 PM, Tuesday 17th of July

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our fifth meeting we will have an in-depth discussion of Problem 3.7 of “Elements of Causal Inference” (link: https://www.dropbox.com/s/o4345krw428kyld/11283.pdf?dl=0) by Bernhard Schölkopf, Jonas Peters and Dominik Janzing.

Week 4

Time: 10AM, Tuesday 10th of July

Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our fourth meeting we will play with some hands on problems at the end of chapter 3 and briefly discuss the assumptions for causal inference in Chapter 2 of “Elements of Causal Inference” (link: https://www.dropbox.com/s/o4345krw428kyld/11283.pdf?dl=0) by Bernhard Schölkopf, Jonas Peters and Dominik Janzing.

Week 3

Time: 11AM, MONDAY 2nd of July Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our third meeting we will the assumption for causal inference in Chapter 2 of “Elements of Causal Inference” (link: https://www.dropbox.com/s/o4345krw428kyld/11283.pdf?dl=0) by Bernhard Schölkopf, Jonas Peters and Dominik Janzing. The book is a machine learning approach to causal inference.

UPDATE: What we ended up doing: Discuss the Eye desease example from chapter 3

Week 2

Time: 10am, Tuesday 26th of June Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

In our second meeting we will discuss the causality principles and its two examples (Section 1.3 and 1.4) in the introduction of “Elements of Causal Inference” (link: https://www.dropbox.com/s/o4345krw428kyld/11283.pdf?dl=0) by Bernhard Schölkopf, Jonas Peters and Dominik Janzing. The book is a machine learning approach to causal inference.

Week 1

Time: 10am, Tuesday 19th of June Location: 4-206A (LSC, Computer Science Department, glass room out of elevator on the left)

The Causality Reading Group will meet for its first meeting this summer. We will discuss the last page of the short 4-page paper on “Statistics for big data: A perspective” by Peter Bühlmannand and Sara van de Geer, ETH Zürich. (attached, link: https://www.sciencedirect.com/science/article/pii/S0167715218300610) Over the course of the summer, we will discuss questions such as: What is causality? How can we identify causality? Can we justify causal statements? Papers, opinions and speakers will come from Philosophy, Computer Science and Mathematics.