Identifying Causal Structure in Dynamical Systems (2020-06-06T00:00:00.000000Z)

TL;DR

The method reveals which state and input variables of the system have a causal influence on each other, and shows that the obtained knowledge of the causal structure reduces the complexity of model learning and yields improved generalization capabilities.

Abstract

We present a method for automatically identifying the causal structure of a dynamical control system. Through a suitable experiment design and subsequent causal analysis, the method reveals, which state and input variables of the system have a causal influence on each other. The experiment design builds on the concept of controllability, which provides a systematic way to compute input trajectories that steer the system to specific regions in its state space. For the causal analysis, we leverage powerful techniques from causal inference and extend them to control systems. Further, we derive conditions that guarantee discovery of the true causal structure of the system and show that the obtained knowledge of the causal structure reduces the complexity of model learning and yields improved generalization capabilities. Experiments on a robot arm demonstrate reliable causal identification from real-world data and extrapolation to regions outside the training domain.

Authors

Sebastian Trimpe

2 papers

Dominik Baumann

1 papers

Friedrich Solowjow

1 papers

TL;DR

Abstract

Authors

References80 items

Probabilistic and Causal Inference: The Works of Judea Pearl

The o80 C++ templated toolbox: Designing customized Python APIs for synchronizing realtime processes

Higher Order Kernel Mean Embeddings to Capture Filtrations of Stochastic Processes

PySINDy: A Python package for the sparse identification of nonlinear dynamical systems from data

A Kernel Two-sample Test for Dynamical Systems

Causal Models for Dynamical Systems

Causality for Machine Learning

Inferring causation from time series in Earth system sciences

Nonlinear System Identification: A User-Oriented Road Map

Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead

Learning stable and predictive structures in kinetic systems

Causal Structure Learning from Multivariate Time Series in Settings with Unmeasured Confounding

The Book of Why: The New Science of Cause and Effect

Markov equivalence of marginalized local independence graphs

Learning Without Mixing: Towards A Sharp Analysis of Linear System Identification

Characterizing and Learning Equivalence Classes of Causal DAGs under Interventions

Theoretical Impediments to Machine Learning With Seven Sparks from the Causal Revolution

Elements of Causal Inference: Foundations and Learning Algorithms

Characteristic and Universal Tensor Product Kernels

Real-Time Perception Meets Reactive Motion Generation

Detecting and quantifying causal associations in large nonlinear time series datasets

From Deterministic ODEs to Dynamic Structural Causal Models

Learning Causal Graphs with Small Interventions

Discovering governing equations from data by sparse identification of nonlinear dynamical systems

Towards a Learning Theory of Cause-Effect Inference

Stochastic Processes and Applications: Diffusion Processes, the Fokker-Planck and Langevin Equations

A Wild Bootstrap for Degenerate Kernel Tests

Causal Discovery via Reproducing Kernel Hilbert Space Embeddings

A Kernel Independence Test for Random Processes

Causal Inference on Time Series using Restricted Structural Equation Models

Identification of dynamic models in complex networks with prediction error methods - Basic methods for consistent module estimates

Topology Identification of Directed Dynamical Networks via Power Spectral Analysis

From Ordinary Differential Equations to Structural Causal Models: the deterministic case

Causal interpretation of stochastic differential equations

Causal inference in time series analysis

A Kernel Two-Sample Test

Stochastic controllability of nonlinear systems

Model learning for robot control: a survey

Estimating the topology of complex dynamical networks by steady state control: Generality and limitation

Estimating the directed information to infer causal relationships in ensemble neural spike train recordings

Graphical Gaussian modelling of multivariate time series with latent variables

Universality, Characteristic Kernels and RKHS Embedding of Measures

Granger Causality and Dynamic Structural Systems

Ten simple rules for dynamic causal modeling

Causal Search in Structural Vector Autoregressive Models

Linear Systems Theory

Structure identification of uncertain general complex dynamical networks with time delay

Hilbert Space Embeddings and Metrics on Probability Measures

Distilling Free-Form Natural Laws from Experimental Data

Search for Additive Nonlinear Time Series Causal Models

Topological identification in networks of dynamical systems

Kernel Measures of Conditional Dependence

Automated reverse engineering of nonlinear dynamical systems

A Candidate to Replace PID Control: SISO-Constrained LQ Control

Causation, Prediction, and Search

Dynamic causal modelling

Searching for the Causal Structure of a Vector Autoregression

The quadruple-tank process: a multivariable laboratory process with an adjustable zero

On Controllability Conception for Stochastic Systems

On Information and Sufficiency

Causal diagrams for empirical research

Optimal control: linear quadratic methods

Dynamic modelling and causality

Deadbeat control and tracking of discrete-time systems

Estimating the Dimension of a Model

Composable Markov processes

On the general theory of control systems

Hybrid System Identification: Theory and Algorithms for Learning Switching Models

Complex Networks: Structure and Dynamics

On the empirical estimation of integral probability metrics

On Causal Discovery from Time Series Data using FCI

Gaussian Processes for Machine Learning, volume 2. MIT press

Identification of piecewise affine systems via mixed-integer programming

Identi cation of piecewise a ne systems via mixed-integer programming

On the Influence of the Kernel on the Consistency of Support Vector Machines

System Identiﬁcation: Theory for the User