Causal Reasoning Workshop 2019

CReW 2019


Probability & Statistics with Applications



The formal study of causality started about 300 hundred years ago with the works of the great philosophers David Hume and Immanuel Kant. Hume approached causality from an empirical perspective: the knowledge that a certain thing (the potential cause) causes or prevents another (the effect) is acquired by means of experience and without any prior knowledge. He identified 3 conditions for this to happen: temporal and spatial contiguity, precedence in time of the cause with respect to the effect and constant conjunction; i.e., the constant occurrence of the both of them. The notion of covariation summarizes such a perspective. Kant, on the other hand, focused on the notion of causal power, which refers to the knowledge that some mechanism or power can cause a certain effect. As can be inferred, this is what we refer to as prior knowledge. Of course, both approaches have their respective advantages and disadvantages. Although the former have given us important clues for capturing the essential features of causality, the latter represent a serious challenge to overcome. Since then, different disciplines such as Philosophy, Psychology, Statistics and Artificial Intelligence (AI) have studied this important phenomenon. One of AI’s main interest is to build intelligent systems capable of automatically acquiring cause-effect relationships and using causal knowledge to build better intelligent systems.
This workshop aims at bringing together researchers and students from Artificial Intelligence, Statistics and Cognitive Science who work on causal theory, the construction of causal models and their evaluation, as well in learning causal models from data. We welcome original research as well as work in progress in the following (but not limited) topics:
Theoretical models of causation
Structural Equation Models
Causal Bayesian networks and other graphical causal models
Metrics and benchmarks for assessing causality
Causal discovery
Causal inference
Machine learning and causality
Applications of causal models
Contributions should be formatted according to Springer LNCS format. We accept 2 types of contributions
Short (6 pgs) - Intended for early research and preliminary results
Long (12 pgs) - Intended for solid pieces of research.
Submissions will be made through EasyChair.