핵심 개념
A kernel-based framework is proposed to model and identify time-invariant systems with fading memory properties, bypassing state-space representations.
초록
The paper introduces a novel approach using kernels to directly model the memory functional of fading memory systems. It explores the benefits of this method in encoding input-output properties and imposing incremental small gain through regularization. The LPV kernel is highlighted for modeling linear parameter varying systems. Examples illustrate the effectiveness of the LPV kernel in system identification, including applications to neuronal circuits like Hodgkin-Huxley models.
통계
"The rapid development of machine learning has boosted our ability to identify nonlinear systems." - 1 sentence.
"The recent article introduced a kernel-based framework for system identification which allows the data fitting to be regularized with input-output properties specified in the form of incremental integral quadratic constraints." - 1 sentence.
"For all δ > 0, there exists ε > 0, such that, for all u, v ∈ Upast sup t≤0 |u(t) − v(t)| < ε ⇒ |Fu − Fv| < δ" - 1 sentence.
"If β2 < 1/c^2, it follows that the operator G has incremental small gain." - 1 sentence.
인용구
"The LPV kernel captures the fading memory properties of circuit components effectively."
"The potential of this framework for identifying non-fading memory systems like Hodgkin-Huxley models is promising."