System Identi cation

By Lennart Ljung. 2014

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Model-based reinforcement learning with parametrized physical models and optimism-driven exploration (2015, methodology): “One way to tackle this challenge is to ﬁrst perform model identiﬁcation, and then use the identiﬁed model to control the system [16], [20].”
Multistep Neural Networks for Data-driven Discovery of Nonlinear Dynamical Systems (2018, background): “The answer to this question pertains to the well-established ﬁeld of systems identiﬁcation [5].”
Non-asymptotic Identification of LTI Systems from a Single Trajectory (2018, background): “Classical results in system identiﬁcation provide asymptotic convergence guarantees for learning models from data [19,29].”
Probabilistic Recurrent State-Space Models (2018, methodology): “System identiﬁcation, i.e. learning dynamics models from data (Ljung, 1998; 2010), is a key ingredient of model-predictive control (Camacho & Alba, 2013) and model-based reinforcement learning (RL) (Deisenroth & Ras-mussen, 2011; Doerr et al., 2017b).”
GNSS Position Integrity in Urban Environments: A Review of Literature (2018, unknown): “The WLS estimator is the best linear unbiased estimator which reaches the Cramer-Rao lower bound [127].”
Exact Recovery of Chaotic Systems from Highly Corrupted Data (2016, background): “Examples include statistical learning theory [Vap13], manifold learning [RS00], machine learning [JM15], physical modeling [Sch13], and system identiﬁcation [Lju98, Lju10].”
Finite-Data Performance Guarantees for the Output-Feedback Control of an Unknown System (2018, result): “…are not only in stark contrast to classical system identiﬁcation results, which typically only provide asymptotic guarantees of model ﬁdelity (see [10] for an overview), but also necessary for the principled integration of these techniques with robust control, as they allow us to quantify the…”
Visual Grounding of Learned Physical Models (2020, unknown): “Prior works on system identi-ﬁcation for robotics usually make strong assumptions on the structure of the underlying system (Ljung, 2001), or require a lot of data to train a forward model (Finn & Levine, 2017), and therefore have a hard time modeling complex de-formable objects like ﬂuids and…”
Gradient Descent Learns Linear Dynamical Systems (2016, methodology): “Maximum likelihood estimation with gradient descent is a popular heuristic for dynamical system identi-ﬁcation [Lju98].”
Symplectic ODE-Net: Learning Hamiltonian Dynamics with Control (2019, unknown): “Such approaches, in contrast to more rigid parametric system identiﬁcation techniques (S¨oderstr¨om & Stoica, 1988), use neural networks to approximate the state-transition dynamics and therefore are more expressive.”
A Broader View on Bias in Automated Decision-Making: Reflecting on Epistemology and Dynamics (2018, methodology): “…the ﬁeld of System Identiﬁcation uses statistical methods to build mathematical models of dynamical systems from measured data, often to be employed to control dynamical systems with strict safety requirements, such as airplanes or electric power systems (Ljung, 1998; ˚Astr¨om & Murray, 2010).”
A Tour of Reinforcement Learning: The View from Continuous Control (2018, background): “Estimation of dynamical systems is called system identiﬁcation in the control community [48].”