Fast Parameter Inference in Nonlinear Dynamical Systems using Iterative Gradient Matching

Authors: Mu Niu, Simon Rogers, Maurizio Filippone, Dirk Husmeier

ICML 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental An evaluation of the proposed method on various benchmark data suggests that it compares favourably with state-of-the-art alternatives.
Researcher Affiliation Academia Mu Niu MU.NIU@GLASGOW.AC.UK School of Mathematics and Statistics, University of Glasgow, UK Simon Rogers SIMON.ROGERS@GLASGOW.AC.UK Department of Computing Science, University of Glasgow,UK Maurizio Filippone MAURIZIO.FILIPPONE@EURECOM.FR Eurecom, France Dirk Husmeier DIRK.HUSMEIER@GLASGOW.AC.UK School of Mathematics and Statistics, University of Glasgow, UK
Pseudocode No The paper describes a multi-step iterative algorithm verbally and in text, but it does not provide a formal pseudocode block or algorithm listing.
Open Source Code No For the GON method, we used the authors software, for RKG2 and RKG3 we used our own code, which is available upon request.
Open Datasets Yes We have evaluated the methods on data generated from two ODE systems: the classical Lotka-Volterra system, and a mathematical description of a protein signal transduction pathway. The Lotka-Volterra equations describe the dynamics of ecological systems with predator-prey interactions (Lotka, 1920). A model for the interactions of five protein isoforms...was studied by Vyshemirsky and Girolami (2008).
Dataset Splits Yes The regularisation parameters λs are estimated using 10-fold cross validation (parallelised!).
Hardware Specification No The paper does not provide specific details about the hardware (e.g., CPU, GPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions using 'the authors software' for the GON method and 'our own code' for RKG2 and RKG3, but it does not list any specific software dependencies or libraries with version numbers.
Experiment Setup Yes A numerical solution of the Lotka-Volterra system gives stationary oscillations, and we therefore chose a stationary kernel: the RBF kernel with state-specific lengthscale ls: ks(tk, ti) = exp l 2 s (tk ti)2 (21). The protein concentrations obtained from the protein transduction pathway are nonstationary...and we have therefore chosen a non-stationary kernel: the multi-layer perceptron kernel (MLP), with state-specific parameters ws and ls, given by k(tk, ti) = arcsin wtkti + l p wt2 i + l + 1 p wt2 k + l + 1 (22).