Multivariate time series are a common data modality in many scientific fields, such as physics, genomics, neuroscience or economics. A major goal in these disciplines is to answer causal questions, such as gene regulatory network inference in biology or identifying causal drivers of extreme (compound) weather events in climate science, which requires the inference of causal relationships in dynamical systems. Similar to the static setting, accessibility, feasibility, and ethical considerations complicate direct experimentation and measurement of such relations in time-dependent settings. While certain properties of time series may aid practical causal inference, such as the arrow of time, most identifiability results for causal discovery as well as cause effect estimation are restricted to the static setting (as in structural causal models or potential outcomes). These results typically do not extend to temporally evolving dynamics. Consequently, natural sciences mostly resort to continuous-time dynamical systems such as ODEs and PDEs to describe hypothesized mechanisms underlying the data generating processes. While such models also allow for causal interpretations as well as predictions under hypothetical interventions, data-driven discovery of continuous-time dynamical systems remains a challenging and relatively underexplored area of research.
In this workshop, we bring together researchers in dynamical systems, time-series methods, causality, infinite-depth neural networks, and machine learning. We believe a side-by-side discussion of dynamical systems and causal inference (discovery and estimation) will allow to develop novel approaches, transfer expertise across communities, and enable us to overcome current limitations of each individual perspective. Connections to other scientific disciplines as well as practitioners’ view will be highlighted to showcase successful applications of causal inference in dynamical settings.
We welcome any contributions on ongoing research at the interface of causality and dynamical systems, including but not limited to:
Causal discovery in time-varying dynamical settings
Identification and estimation of causal effects in dynamical systems
Granger causality
Data-driven discovery of dynamical laws (e.g. symbolic regression)
Deep learning for time-series modeling and prediction (e.g. neural ODEs, physics-inspired NNs)
Applications of dynamical modeling and causal inference for scientific discovery
Paper submission deadline: September 26 11:59 pm AoE, 2022
Notification to authors: October 20, 2022
Camera-ready version: TBA
Workshop date: December 3, 2022
We invite submissions on on-going research that have not yet been published in a venue with proceedings. While we welcome unfinished work, submissions in this track should contain original ideas, connections, or results. The main body of the submission (including most important figures and tables) must not exceed 9 pages with unlimited additional space for references and supplementary material. Reviewers will be asked to judge the main body of the paper and will not be required to read the supplementary material. Papers should be submitted anonymously to the OpenReview submission portal as a single pdf formatted using the latex style file provided below. All papers will undergo double-blind peer review based on which a subset will be invited for a contributed talk; all other accepted papers will be invited to be presented at the poster sessions. The workshop will not have proceedings.