Call for Papers

Causality is central to a broad range of disciplines, including human cognition, engineering, and scientific discovery. It’s pervasive in human cognition, shaping how we view, understand, and react to the world around us. It’s a key ingredient in building AI systems that are autonomous and can act efficiently in complex and uncertain environments. It’s also important to the process of scientific discovery since it form underpins the scientific method and how explanations are constructed.

Not surprisingly, the tasks of learning and reasoning with causal-effect relationships have attracted great interest in the artificial intelligence and machine learning communities. This effort has led to a very general theoretical and algorithmic understanding of what causality means and under what conditions it can be inferred. These results have started to percolate through more applied fields that generate the bulk of the data currently available, ranging from genetics to medicine, from psychology to economics.

Here is a list of typical research contributions that would have a high probability of being appreciated (not exhaustive):

Causal inference and discovery in complex environments (e.g., from noisy, high-dimensional, and biased data);
Learning with heterogenous datasets;
Measures and methods for evaluating the quality of causal predictions;
Addressing the challenge of practical causal inference in the context of real applications;
Real-world validation of methods for causal inference and discovery;
Causal predictions in complex (e.g., heterogeneous, confounded, and cyclic) settings;
Occam’s Razor in causal inference;
Relations between causal modeling and data science.

Example Topics

The community currently spans through several research directions, and some of the most recent focus includes, but are not limited to, new theory and methods for:

Causal inference by combining data from heterogeneous sources, such as the fusion of experimental and observational data, and issues of transportability, selection bias, and missing data, as well as sets of overlapping but different measurements;
Structure learning with latent variables exploiting different types of constraints, including conditional independencies, semiparametric and parametric constraints, and functional and inequalities constraints;
Optimal design of experiments and interventions needed to discriminate between causal structures compatible with the current observed data;
Applications in Machine learning (e.g., reinforcement learning, transfer learning, recommender systems) and in the empirical sciences (biology, medicine, economics, social sciences).

Submission

There are two possible submission formats. The authors can either submit:

a full-length paper, limited to 9 pages (including figures and text, excluding references), or
a one-page summary (excluding references) describing an open problem (see below).

Full Papers

If a contribution consists of material that has been published elsewhere earlier, the authors must indicate this in their submission and cite the original work. We will give preference to original work.

Our submission deadline comes a few days after the UAI author notification deadline. We encourage co-submission of (full) papers that have been submitted to the main UAI 2017 conference. Please indicate if your paper was also submitted to UAI. If accepted for UAI, the paper will be published in the UAI proceedings, but we may also invite the authors to give a (oral or poster) presentation at the workshop.

Style files for full papers can be found on the UAI website. As with the main conference, please prepare your paper for double-blind review, i.e. without author information in the paper itself.

Open Problems

This year we will try to include a session on Open Problems in Causality. We encourage submissions of succinct descriptions (one page) of the problem and will invite authors to give a short presentation at the workshop, followed by discussion. The Open Problems proposals will only be reviewed by the workshop committee in order to avoid presentations on problems that are ill-posed or obviously resolved. We very much hope that there will be interest in such a session, not just as an audience member, but also as a contributor.

Logistics

Open problem proposals and full papers must be submitted via e-mail by the deadline (June 17 and May 19, respectively) to uai2017.causalityworkshop@gmail.com.

Contributions will be peer reviewed by at least two reviewers. Accepted papers will be presented either as oral presentation or in a poster session.

Proceedings

After the workshop, we will publish proceedings via this web-page and on a special issue in a reputable journal, assuming we can satisfy their length criteria.

Authors of accepted papers can choose to contribute the submitted manuscript. They can also choose not to contribute to the proceedings.

Oral presentation slides will also be disseminated via this web-page.

Important Dates

May 26 ~~May 19~~	paper submission deadline
June 26 ~~June 16~~	author notification
June 17	open problem submission deadline
June 26	author notification for open problem submission
July 15	workshop paper version due
August 15	Workshop (last day of the UAI 2017 main conference, August 11-15)